lstm_time_variant_channel.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "source": [
    "<a href=\"https://colab.research.google.com/github/lasseufpa/ml4comm/blob/main/lstm_time_variant_channel.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
   ],
   "metadata": {}
  },
  {
   "cell_type": "markdown",
   "source": [
    "Globecom Tutorial - December 7, 2021\r\n",
    "Tutorial 29: Machine Learning for MIMO Systems with Large Arrays\r\n",
    "Nuria Gonzalez-Prelcic (NCSU),\r\n",
    "Aldebaro Klautau (UFPA) and\r\n",
    "Robert W. Heath Jr. (NCSU)"
   ],
   "metadata": {}
  },
  {
   "cell_type": "markdown",
   "source": [
    "# LSTM with time variant channel"
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "source": [
    "# Clone the repository if running in Colab and install all the dependencies\r\n",
    "if 'google.colab' in str(get_ipython()):\r\n",
    "    import sys\r\n",
    "    import os\r\n",
    "    try:\r\n",
    "      !git clone https://github.com/lasseufpa/ml4comm.git\r\n",
    "    except:\r\n",
    "      print(\"ml4comm is already in the contents\")\r\n",
    "    !pip install -r /content/ml4comm/requirements.txt &>/dev/null\r\n",
    "    sys.path.append(os.path.abspath('/content/ml4comm'))"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "source": [
    "import sys\r\n",
    "import pandas as pd\r\n",
    "import numpy as np\r\n",
    "import matplotlib.pyplot as plt\r\n",
    "import matplotlib.cm as cmap\r\n",
    "import commpy.modulation as cm\r\n",
    "import commpy.utilities as cu\r\n",
    "from numpy.random import randn, random, standard_normal\r\n",
    "import tensorflow as tf\r\n",
    "import seaborn as sns\r\n",
    "from sklearn.metrics import confusion_matrix, accuracy_score\r\n",
    "from sklearn.tree import DecisionTreeClassifier\r\n",
    "from tensorflow.keras.models import Sequential, load_model\r\n",
    "from tensorflow.keras.layers import (\r\n",
    "    Dense,\r\n",
    "    LSTM,\r\n",
    "    Dropout,\r\n",
    "    Lambda,\r\n",
    "    Flatten,\r\n",
    "    TimeDistributed,\r\n",
    ")\r\n",
    "from tensorflow.keras.optimizers import Adam, SGD\r\n",
    "from sklearn.preprocessing import MinMaxScaler\r\n",
    "\r\n",
    "from ml4comm.qam_analyzer import plot_decision_boundary, ser"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Time variant channel problem\n",
    "Consider a situation where a crazy channel has a cyclic behavior in time, where each time a symbol is sent through the channel the channel has one of 5 possible behaviors. This causes an interesting case, where a conventional classifier can get confused as it probably will not understand the cyclic behavior and therefore will make wrong class predictions.\n",
    "\n",
    "This way, in this case, a model that can understand the time dependencies of the input can learn the cyclic behavior and predict the correct class as output. In this sense, the long short-term memory(LSTM) model was chosen, as it is prepared to deal with sequence inputs and understand the temporal relationship between them.  "
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "source": [
    "def generate_symbols(transmissions=100, M=16):\n",
    "    \"\"\"\n",
    "    Parameters\n",
    "    ----------\n",
    "    transmissions: int\n",
    "        Number of transmissions. Default is 100.\n",
    "    M: int\n",
    "        Number of symbols in the constellation. Default is 16.\n",
    "    Returns\n",
    "    -------\n",
    "    \"\"\"\n",
    "    bits_per_QAMsymbol = int(np.log2(M))\n",
    "    bitarrays = [cu.dec2bitarray(obj, bits_per_QAMsymbol)\n",
    "                 for obj\n",
    "                 in np.arange(0, M)]\n",
    "    #print(bitarrays)\n",
    "    qammod = cm.QAMModem(M)\n",
    "    const = []\n",
    "    for i in range(len(bitarrays)):\n",
    "        const.append(qammod.modulate(bitarrays[i]))\n",
    "    # unit average power\n",
    "    constellation = const / np.sqrt((M - 1) * (2 ** 2) / 6)\n",
    "\n",
    "    ind = np.random.randint(M, size=[1, transmissions])\n",
    "\n",
    "    # QAM symbols for each antenna\n",
    "    x = constellation[ind]\n",
    "    #Reshape to (num_elements,) instead of (num_elements,1)\n",
    "    #https://www.semicolonworld.com/question/59886/numpy-vector-n-1-dimension-gt-n-dimension-conversion\n",
    "    x = x.reshape((-1,))\n",
    "    ind = ind.reshape((-1,))\n",
    "    return x, ind\n"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "source": [
    "def variable_crazy_channel_propagate(x, n_patterns=5):\n",
    "    \"\"\"\n",
    "    Parameters\n",
    "    ----------\n",
    "    x:\n",
    "        Array of symbols to be propagated in the time variable channel.\n",
    "    n_patterns:\n",
    "        Number of time variable patterns\n",
    "    Returns\n",
    "    -------\n",
    "    y:\n",
    "        Channel output\n",
    "    noise:\n",
    "        Noise added to the channel input\n",
    "    \"\"\"\n",
    "    y = np.zeros(x.shape, dtype=np.complex_)\n",
    "    num_samples = x.shape[0]\n",
    "    # Noise parametter depending in the amount of patterns\n",
    "    n_params = {'std':           np.random.uniform( low=0.1, high=0.125, size=n_patterns),\n",
    "                'pwr_factor':    np.random.uniform( low=0.002, high=0.0035, size=n_patterns),\n",
    "                'i_mag':    np.random.uniform( low=-1, high=1, size=n_patterns),\n",
    "                'r_mag':    np.random.uniform( low=-1, high=1, size=n_patterns)}\n",
    "\n",
    "    # The std and pwr_factor affect snr, while i_mag affects the disposition of the symbols\n",
    "    for i in range(x.shape[1]):\n",
    "        gaussian_noise = (n_params['r_mag'][i]*standard_normal(num_samples) \\\n",
    "                            + n_params['i_mag'][i]*1j * standard_normal(num_samples)) \\\n",
    "                            * n_params['std'][i] * 0.5\n",
    "        noise = np.abs(x[:,i])*n_params['pwr_factor'][i] + gaussian_noise\n",
    "\n",
    "        if i == 0:\n",
    "            y[:,i] = x[:,i] * noise + 1/x[:,i]\n",
    "        elif i == 1:\n",
    "            y[:,i] = (x[:,i] * noise + 1/x[:,i]) * (np.cos(1.25663706144*i)+1j*np.sin(1.25663706144*i))\n",
    "        elif i == 2:\n",
    "            y[:,i] = (x[:,i] * noise + 1/x[:,i]) * (np.cos(1.25663706144*i)+1j*np.sin(1.25663706144*i))\n",
    "        elif i == 3:\n",
    "            y[:,i] = x[:,i] + (noise * 1/x[:,i])\n",
    "        elif i == 4:\n",
    "            y[:,i] = (x[:,i] * noise - 1/x[:,i]) * (np.cos(np.pi/2)+1j*np.sin(np.pi/2))\n",
    "    \n",
    "    return np.array(y).flatten()"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "source": [
    "n_patterns = 5\n",
    "M = 16\n",
    "num_of_symbols = 1000*n_patterns\n",
    "symbs, indices=generate_symbols(transmissions=num_of_symbols, M=M)\n",
    "symbs = np.reshape(symbs, (int(len(symbs)/n_patterns),n_patterns))\n",
    "\n",
    "#transmit over the channel\n",
    "channel_output = variable_crazy_channel_propagate(symbs, n_patterns)"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "source": [
    "# Generate dataset\n",
    "# Train\n",
    "train_size = int(0.5*len(indices))\n",
    "y_train = indices[:train_size]\n",
    "X_train = np.stack([np.real(channel_output[:train_size]),\n",
    "                    np.imag(channel_output[:train_size])], axis=1)\n",
    "\n",
    "# Test\n",
    "y_test = indices[train_size:]\n",
    "X_test = np.stack([np.real(channel_output[train_size:]),\n",
    "                   np.imag(channel_output[train_size:])], axis=1)"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Lets have a glimpse of the channel output"
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "source": [
    "def plot_symbols(ch_out_symbs, true_symbs, indices, num_classes):\n",
    "    custom_cmap = cmap.get_cmap('tab20')\n",
    "    colors = custom_cmap.colors[:num_classes]\n",
    "    levels = np.arange(num_classes + 2) - 0.5\n",
    "    y_train = indices[:len(ch_out_symbs)]\n",
    "    for ii in range(num_classes):\n",
    "        selected_indices = np.argwhere(y_train==ii)\n",
    "        selected_indices = selected_indices.reshape((-1))\n",
    "        plt.plot(ch_out_symbs[selected_indices, 0], ch_out_symbs[selected_indices, 1], 'o', color=colors[ii], label=f'{ii}')\n",
    "        \n",
    "    plt.plot(np.real(true_symbs), np.imag(true_symbs), 'rx')\n",
    "    plt.legend(title='Classes', bbox_to_anchor=(1, 1), loc='upper left', ncol=2, handleheight=2, labelspacing=0.05, frameon=False)\n",
    "    plt.ylabel('Quadrature')\n",
    "    plt.xlabel('In-phase')\n",
    "\n",
    "    plt.show()  "
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "source": [
    "plot_symbols(X_train, symbs, indices, M)"
   ],
   "outputs": [
    {
     "output_type": "display_data",
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     }
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "markdown",
   "source": [
    "## There is 5 different patterns in this output, these patterns are cyclic in a way we can divide the original channel output into 5 distinct behaviors"
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "source": [
    "channel_output_reshape = np.reshape(channel_output, (int(len(channel_output)/n_patterns),n_patterns))\n",
    "indices_reshape = np.reshape(indices, (int(len(indices)/n_patterns),n_patterns))"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "source": [
    "channel_output_slice = np.stack([np.real(channel_output_reshape[:,0][:train_size]),\n",
    "                     np.imag(channel_output_reshape[:,0][:train_size])], axis=1)\n",
    "plot_symbols(channel_output_slice, symbs, indices_reshape[:,0], M)"
   ],
   "outputs": [
    {
     "output_type": "display_data",
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     }
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "source": [
    "channel_output_slice = np.stack([np.real(channel_output_reshape[:,1][:train_size]),\n",
    "                     np.imag(channel_output_reshape[:,1][:train_size])], axis=1)\n",
    "plot_symbols(channel_output_slice, symbs, indices_reshape[:,1], M)"
   ],
   "outputs": [
    {
     "output_type": "display_data",
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     }
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "source": [
    "channel_output_slice = np.stack([np.real(channel_output_reshape[:,2][:train_size]),\n",
    "                     np.imag(channel_output_reshape[:,2][:train_size])], axis=1)\n",
    "plot_symbols(channel_output_slice, symbs, indices_reshape[:,2], M)"
   ],
   "outputs": [
    {
     "output_type": "display_data",
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     }
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "source": [
    "channel_output_slice = np.stack([np.real(channel_output_reshape[:,3][:train_size]),\n",
    "                     np.imag(channel_output_reshape[:,3][:train_size])], axis=1)\n",
    "plot_symbols(channel_output_slice, symbs, indices_reshape[:,3], M)"
   ],
   "outputs": [
    {
     "output_type": "display_data",
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     }
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "source": [
    "channel_output_slice = np.stack([np.real(channel_output_reshape[:,4][:train_size]),\n",
    "                     np.imag(channel_output_reshape[:,4][:train_size])], axis=1)\n",
    "plot_symbols(channel_output_slice, symbs, indices_reshape[:,4], M)"
   ],
   "outputs": [
    {
     "output_type": "display_data",
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     }
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Now let us try to train a model to predict the symbol desired location without the notion of time dependence behavior"
   ],
   "metadata": {}
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Decision Tree"
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "source": [
    "class_DT = DecisionTreeClassifier(max_depth=20, random_state=42)\n",
    "class_DT.fit(X_train, y_train)\n",
    "plot_decision_boundary(class_DT, X_train, y_train)"
   ],
   "outputs": [
    {
     "output_type": "display_data",
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     }
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "source": [
    "y_pred = class_DT.predict(X_test)\n",
    "conf_mtx = confusion_matrix(y_test, y_pred)\n",
    "\n",
    "conf_df = pd.DataFrame(conf_mtx, index = [i for i in range(M)],\n",
    "                  columns = [i for i in range(M)])\n",
    "plt.figure(figsize = (10,7))\n",
    "sns.heatmap(conf_df, cmap=sns.cm.rocket_r, square=True, linewidths=0.1, annot=True, fmt='d', annot_kws={\"fontsize\": 8})  \n",
    "plt.show()"
   ],
   "outputs": [
    {
     "output_type": "display_data",
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 720x504 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     }
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "source": [
    "ser = ser(class_DT, X_test, y_test)\n",
    "print(f'Decision Tree:\\n SER:\\t {ser:.3f}')\n",
    "\n",
    "acc = accuracy_score(y_test, y_pred)\n",
    "print(f' Acc:\\t {acc:.3f}')"
   ],
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "Decision Tree:\n",
      " SER:\t 0.201\n",
      " Acc:\t 0.799\n"
     ]
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Now let us fit a model that has the notion of time dependency\n",
    "\n",
    "The model chosen to do so is the long short-term memory(LSTM), with the help of recurrence, this model can take multiple symbols as entry and predict the last symbol class. This way the model can learn the cyclic behavior produced by the variable channel and potentially learn each cycle. "
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "source": [
    "features = np.stack([np.real(channel_output),\n",
    "                    np.imag(channel_output)], axis=1)"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "markdown",
   "source": [
    "### First let us generate rolling windows of symbols as the features to our LSTM model"
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "source": [
    "def rolling_window_mtx(x, window_size):\n",
    "        \"\"\"Compute all overlapping (rolling) observation windows over a vector \n",
    "            and return a matrix\n",
    "\n",
    "        Args:\n",
    "            x           : observation vector that is supposed to be split into\n",
    "                          overlapping windows\n",
    "            window_size : the target window size\n",
    "\n",
    "        Returns:\n",
    "\n",
    "            Window matrix with all windows as rows. That is, if n_windows is the\n",
    "            number of windows, the result has dimensions:\n",
    "\n",
    "            (n_windows, window_size)\n",
    "\n",
    "        \"\"\"\n",
    "        if window_size < 1:\n",
    "            raise ValueError(\"`window_size` must be at least 1.\")\n",
    "        if window_size > x.shape[-1]:\n",
    "            raise ValueError(\"`window_size` is too long.\")\n",
    "\n",
    "        shape = x.shape[:-1] + (x.shape[-1] - window_size + 1, window_size)\n",
    "        strides = x.strides + (x.strides[-1],)\n",
    "\n",
    "        return np.lib.stride_tricks.as_strided(x, shape=shape, strides=strides)"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "source": [
    "n_time_steps = 18 # window size\n",
    "scaler = MinMaxScaler((-1,1)) # Scale the symbol real and imaginary part\n",
    "scaler.fit(features)\n",
    "features = scaler.transform(features)\n",
    "\n",
    "n_windows   = features.shape[0] - n_time_steps - 1 # Total number of Rolling windows\n",
    "feature_win = 2 * n_time_steps # Real and imaginnary coordinates of 5 instances\n",
    "feature_mtx = np.zeros((n_windows, feature_win))\n",
    "x_train_r = features[:,0]\n",
    "x_train_i = features[:,1]\n",
    "rolling_window_mtx( x_train_r[:-1], n_time_steps).shape\n",
    "\n",
    "# Get the real coordinates and allocate to the first set of columns\n",
    "feature_mtx[:, range(0, n_time_steps * 2, 2)] = rolling_window_mtx(\n",
    "     x_train_r[:-1], n_time_steps)[:-1]\n",
    "\n",
    "# Now we have a set of windows of the real coordinate\n",
    "# Lets take a look in one window\n",
    "print(f\"\\n One window containing just the real coordinates : \\n {feature_mtx[0]}\")\n",
    "\n",
    "# Do the same for the imaginary coordinates\n",
    "feature_mtx[:, range(1, n_time_steps * 2, 2)] = rolling_window_mtx(\n",
    "                        x_train_i[:-1], n_time_steps)[:-1]\n",
    "\n",
    "# Lets take a look in the same window\n",
    "print(f\"\\n One window with all the coordinates: \\n {feature_mtx[0]}\")"
   ],
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "\n",
      " One window containing just the real coordinates : \n",
      " [-0.24083467  0.          0.89416995  0.         -0.99585781  0.\n",
      "  0.16453376  0.         -0.22917568  0.          0.7183538   0.\n",
      "  0.9024236   0.         -0.08726271  0.         -0.42448694  0.\n",
      " -0.13081147  0.         -0.25332151  0.          0.14782434  0.\n",
      "  0.26665227  0.          0.43467367  0.          0.43852912  0.\n",
      "  0.42689462  0.         -0.89948154  0.          0.12875982  0.        ]\n",
      "\n",
      " One window with all the coordinates: \n",
      " [-0.24083467  0.23612374  0.89416995  0.4606032  -0.99585781 -0.16488436\n",
      "  0.16453376 -0.20103093 -0.22917568 -0.24807425  0.7183538  -0.71776444\n",
      "  0.9024236   0.45094082 -0.08726271  0.32334609 -0.42448694 -0.42444033\n",
      " -0.13081147 -0.46308502 -0.25332151 -0.24365857  0.14782434 -0.30002059\n",
      "  0.26665227 -0.3692805   0.43467367 -0.43659976  0.43852912 -0.14070176\n",
      "  0.42689462  0.13239232 -0.89948154 -0.4550942   0.12875982  0.42479744]\n"
     ]
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Now we need to reshape this feature matrix  the LSTM network\n",
    "The LSTM network receives as entry feature a matrix of 3 dimensions, these being samples, timestamps, and features. For our problem, we selected 18 timestamps of 2 features (real and imaginary coordinates). This way, the first dimension, the dimension of the sample will contain the timestamps dimension (in this case 18 timestamps) and for last each timestamp will contain 2 features."
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "source": [
    "shape = feature_mtx.shape\n",
    "\n",
    "# Reshape the feature matrix to samples, timestamps, features\n",
    "feature_mtx_3dim = np.reshape(\n",
    "    feature_mtx, (shape[0], n_time_steps, 2))"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Generate the labels of an LSTM network\n",
    "Now for the labels of the LSTM network, not all the cases require a label matrix of 3 dimensions. The cases where this is required are on sequence to sequence problems, where the model is made to predict a sequence of timestamps of one or more features. However for our problem, the LSTM network needs to predict the output class of the last timestamp feature, this way we can use a label matrix of 2 dimensions, these being samples and label class."
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "source": [
    "# Get the labels for the feature matrix\n",
    "label_mtx = rolling_window_mtx(\n",
    "                    indices[n_time_steps-1:], 1)[:-2]"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "markdown",
   "source": [
    "### Now that we have the label matrix containing the output class we can transform it to a categorical form with a one-hot encoding transformation."
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "source": [
    "label_mtx_cat = tf.keras.utils.to_categorical(label_mtx)"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "markdown",
   "source": [
    "### With the dataset well defined we can divide between training and test sets and observe one sample of the training features and labels"
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "source": [
    "size = feature_mtx_3dim.shape[0]\n",
    "lstm_x_train = feature_mtx_3dim[:int(0.5*size)] # Split in 50% the dataset in training and test\n",
    "lstm_x_test  = feature_mtx_3dim[int(0.5*size):]\n",
    "\n",
    "lstm_y_train = label_mtx_cat[:int(0.5*size)]\n",
    "lstm_y_test  = label_mtx_cat[int(0.5*size):]\n",
    "\n",
    "# Lets take a look in the same window\n",
    "print(f\"One sample unit of the LSTM entry features: \\n {lstm_x_train[0]}\")\n",
    "print(f\"LSTM features shape: \\n {lstm_x_train.shape}\")\n",
    "print(f\"\\n One sample unit of the LSTM labels: \\n {lstm_y_train[0]}\")\n",
    "print(f\"LSTM labels shape: \\n {lstm_y_train.shape}\")"
   ],
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "One sample unit of the LSTM entry features: \n",
      " [[-0.24083467  0.23612374]\n",
      " [ 0.89416995  0.4606032 ]\n",
      " [-0.99585781 -0.16488436]\n",
      " [ 0.16453376 -0.20103093]\n",
      " [-0.22917568 -0.24807425]\n",
      " [ 0.7183538  -0.71776444]\n",
      " [ 0.9024236   0.45094082]\n",
      " [-0.08726271  0.32334609]\n",
      " [-0.42448694 -0.42444033]\n",
      " [-0.13081147 -0.46308502]\n",
      " [-0.25332151 -0.24365857]\n",
      " [ 0.14782434 -0.30002059]\n",
      " [ 0.26665227 -0.3692805 ]\n",
      " [ 0.43467367 -0.43659976]\n",
      " [ 0.43852912 -0.14070176]\n",
      " [ 0.42689462  0.13239232]\n",
      " [-0.89948154 -0.4550942 ]\n",
      " [ 0.12875982  0.42479744]]\n",
      "LSTM features shape: \n",
      " (2490, 18, 2)\n",
      "\n",
      " One sample unit of the LSTM labels: \n",
      " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
      "LSTM labels shape: \n",
      " (2490, 16)\n"
     ]
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Define the LSTM"
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "source": [
    "lr=0.01\n",
    "lr_decay=0.0001\n",
    "optimizer = Adam(learning_rate=lr, decay=lr_decay)\n",
    "n_epochs=200\n",
    "n_mini_batch=60\n",
    "\n",
    "n_timesteps, n_features, n_outputs = lstm_x_train.shape[1], lstm_x_train.shape[2], lstm_y_train.shape[1]\n",
    "lstm_categorical = Sequential()\n",
    "lstm_categorical.add(LSTM(100, input_shape=(n_timesteps,n_features), return_sequences=True))\n",
    "lstm_categorical.add(LSTM(100))\n",
    "lstm_categorical.add(Dropout(0.5))\n",
    "lstm_categorical.add(Dense(100, activation='relu'))\n",
    "lstm_categorical.add(Dense(100, activation='relu'))\n",
    "lstm_categorical.add(Dense(n_outputs, activation='softmax'))\n",
    "lstm_categorical.compile(loss='categorical_crossentropy', optimizer=optimizer, metrics=['accuracy'])\n"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Train the model"
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "source": [
    "history = lstm_categorical.fit(\n",
    "            lstm_x_train,\n",
    "            lstm_y_train,\n",
    "            batch_size=n_mini_batch,\n",
    "            epochs=n_epochs,\n",
    "        )"
   ],
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "Epoch 1/200\n",
      "42/42 [==============================] - 4s 42ms/step - loss: 2.7799 - accuracy: 0.0627\n",
      "Epoch 2/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 2.7703 - accuracy: 0.0767\n",
      "Epoch 3/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 2.6936 - accuracy: 0.0920\n",
      "Epoch 4/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 2.6244 - accuracy: 0.1225\n",
      "Epoch 5/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 2.5196 - accuracy: 0.1506\n",
      "Epoch 6/200\n",
      "42/42 [==============================] - 2s 46ms/step - loss: 2.4586 - accuracy: 0.1534\n",
      "Epoch 7/200\n",
      "42/42 [==============================] - 2s 49ms/step - loss: 2.3754 - accuracy: 0.1594\n",
      "Epoch 8/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 2.3225 - accuracy: 0.1896\n",
      "Epoch 9/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 2.2414 - accuracy: 0.2165\n",
      "Epoch 10/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 2.1829 - accuracy: 0.2161\n",
      "Epoch 11/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 2.1164 - accuracy: 0.2418\n",
      "Epoch 12/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 2.0186 - accuracy: 0.2675\n",
      "Epoch 13/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 1.9435 - accuracy: 0.2867\n",
      "Epoch 14/200\n",
      "42/42 [==============================] - 2s 45ms/step - loss: 1.8149 - accuracy: 0.3249\n",
      "Epoch 15/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 1.6481 - accuracy: 0.3570\n",
      "Epoch 16/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 1.6107 - accuracy: 0.3647\n",
      "Epoch 17/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 1.5292 - accuracy: 0.3795\n",
      "Epoch 18/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 1.4892 - accuracy: 0.3912\n",
      "Epoch 19/200\n",
      "42/42 [==============================] - 2s 43ms/step - loss: 1.3259 - accuracy: 0.4414\n",
      "Epoch 20/200\n",
      "42/42 [==============================] - 2s 43ms/step - loss: 1.3383 - accuracy: 0.4526\n",
      "Epoch 21/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 1.2357 - accuracy: 0.4711\n",
      "Epoch 22/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 1.1551 - accuracy: 0.5133\n",
      "Epoch 23/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 1.1543 - accuracy: 0.5012\n",
      "Epoch 24/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 1.1004 - accuracy: 0.5277\n",
      "Epoch 25/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 1.0662 - accuracy: 0.5273\n",
      "Epoch 26/200\n",
      "42/42 [==============================] - 2s 43ms/step - loss: 1.0391 - accuracy: 0.5373\n",
      "Epoch 27/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 1.0015 - accuracy: 0.5610\n",
      "Epoch 28/200\n",
      "42/42 [==============================] - 2s 46ms/step - loss: 0.9842 - accuracy: 0.5627\n",
      "Epoch 29/200\n",
      "42/42 [==============================] - 2s 48ms/step - loss: 0.9559 - accuracy: 0.5675\n",
      "Epoch 30/200\n",
      "42/42 [==============================] - 2s 51ms/step - loss: 0.9311 - accuracy: 0.5759\n",
      "Epoch 31/200\n",
      "42/42 [==============================] - 2s 47ms/step - loss: 0.9129 - accuracy: 0.5775\n",
      "Epoch 32/200\n",
      "42/42 [==============================] - 2s 52ms/step - loss: 0.9024 - accuracy: 0.5920\n",
      "Epoch 33/200\n",
      "42/42 [==============================] - 2s 56ms/step - loss: 0.8673 - accuracy: 0.6032\n",
      "Epoch 34/200\n",
      "42/42 [==============================] - 2s 56ms/step - loss: 0.8450 - accuracy: 0.6056\n",
      "Epoch 35/200\n",
      "42/42 [==============================] - 2s 53ms/step - loss: 0.8248 - accuracy: 0.6084\n",
      "Epoch 36/200\n",
      "42/42 [==============================] - 2s 45ms/step - loss: 0.8515 - accuracy: 0.6257\n",
      "Epoch 37/200\n",
      "42/42 [==============================] - 2s 44ms/step - loss: 0.8058 - accuracy: 0.6309\n",
      "Epoch 38/200\n",
      "42/42 [==============================] - 2s 47ms/step - loss: 0.8142 - accuracy: 0.6245\n",
      "Epoch 39/200\n",
      "42/42 [==============================] - 2s 47ms/step - loss: 0.7829 - accuracy: 0.6273\n",
      "Epoch 40/200\n",
      "42/42 [==============================] - 2s 49ms/step - loss: 0.7771 - accuracy: 0.6353\n",
      "Epoch 41/200\n",
      "42/42 [==============================] - 2s 45ms/step - loss: 0.7394 - accuracy: 0.6418\n",
      "Epoch 42/200\n",
      "42/42 [==============================] - 2s 47ms/step - loss: 0.7967 - accuracy: 0.6261\n",
      "Epoch 43/200\n",
      "42/42 [==============================] - 2s 45ms/step - loss: 0.7640 - accuracy: 0.6357\n",
      "Epoch 44/200\n",
      "42/42 [==============================] - 2s 46ms/step - loss: 0.7443 - accuracy: 0.6438\n",
      "Epoch 45/200\n",
      "42/42 [==============================] - 2s 45ms/step - loss: 0.7220 - accuracy: 0.6498\n",
      "Epoch 46/200\n",
      "42/42 [==============================] - 2s 47ms/step - loss: 0.7382 - accuracy: 0.6450\n",
      "Epoch 47/200\n",
      "42/42 [==============================] - 2s 43ms/step - loss: 0.7611 - accuracy: 0.6386\n",
      "Epoch 48/200\n",
      "42/42 [==============================] - 2s 43ms/step - loss: 0.6923 - accuracy: 0.6735\n",
      "Epoch 49/200\n",
      "42/42 [==============================] - 2s 44ms/step - loss: 0.7178 - accuracy: 0.6675\n",
      "Epoch 50/200\n",
      "42/42 [==============================] - 2s 44ms/step - loss: 0.7210 - accuracy: 0.6602\n",
      "Epoch 51/200\n",
      "42/42 [==============================] - 2s 46ms/step - loss: 0.7141 - accuracy: 0.6574\n",
      "Epoch 52/200\n",
      "42/42 [==============================] - 2s 44ms/step - loss: 0.7290 - accuracy: 0.6442\n",
      "Epoch 53/200\n",
      "42/42 [==============================] - 2s 44ms/step - loss: 0.7038 - accuracy: 0.6687\n",
      "Epoch 54/200\n",
      "42/42 [==============================] - 2s 48ms/step - loss: 0.6937 - accuracy: 0.6695\n",
      "Epoch 55/200\n",
      "42/42 [==============================] - 2s 44ms/step - loss: 0.6858 - accuracy: 0.6743\n",
      "Epoch 56/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.7026 - accuracy: 0.6679\n",
      "Epoch 57/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.6685 - accuracy: 0.6835\n",
      "Epoch 58/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.6693 - accuracy: 0.6799\n",
      "Epoch 59/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.6677 - accuracy: 0.6727\n",
      "Epoch 60/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.6565 - accuracy: 0.6968\n",
      "Epoch 61/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.6715 - accuracy: 0.6904\n",
      "Epoch 62/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.6586 - accuracy: 0.6863\n",
      "Epoch 63/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.6178 - accuracy: 0.7048\n",
      "Epoch 64/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 0.6376 - accuracy: 0.7060\n",
      "Epoch 65/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.6124 - accuracy: 0.7141\n",
      "Epoch 66/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 0.6448 - accuracy: 0.7028\n",
      "Epoch 67/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.6282 - accuracy: 0.7137\n",
      "Epoch 68/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 0.6604 - accuracy: 0.7000\n",
      "Epoch 69/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.6269 - accuracy: 0.7104\n",
      "Epoch 70/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.6003 - accuracy: 0.7341\n",
      "Epoch 71/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.6093 - accuracy: 0.7233\n",
      "Epoch 72/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.5831 - accuracy: 0.7438\n",
      "Epoch 73/200\n",
      "42/42 [==============================] - 2s 43ms/step - loss: 0.5865 - accuracy: 0.7345\n",
      "Epoch 74/200\n",
      "42/42 [==============================] - 2s 43ms/step - loss: 0.5672 - accuracy: 0.7530\n",
      "Epoch 75/200\n",
      "42/42 [==============================] - 2s 43ms/step - loss: 0.5898 - accuracy: 0.7410\n",
      "Epoch 76/200\n",
      "42/42 [==============================] - 2s 44ms/step - loss: 0.5609 - accuracy: 0.7454\n",
      "Epoch 77/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.5430 - accuracy: 0.7566\n",
      "Epoch 78/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.5265 - accuracy: 0.7775\n",
      "Epoch 79/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.5181 - accuracy: 0.8000\n",
      "Epoch 80/200\n",
      "42/42 [==============================] - 2s 47ms/step - loss: 0.5249 - accuracy: 0.7839\n",
      "Epoch 81/200\n",
      "42/42 [==============================] - 2s 45ms/step - loss: 0.4896 - accuracy: 0.8080\n",
      "Epoch 82/200\n"
     ]
    },
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "42/42 [==============================] - 2s 46ms/step - loss: 0.4482 - accuracy: 0.8329\n",
      "Epoch 83/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.4181 - accuracy: 0.8410\n",
      "Epoch 84/200\n",
      "42/42 [==============================] - 2s 44ms/step - loss: 0.3974 - accuracy: 0.8606\n",
      "Epoch 85/200\n",
      "42/42 [==============================] - 2s 44ms/step - loss: 0.3105 - accuracy: 0.8900\n",
      "Epoch 86/200\n",
      "42/42 [==============================] - 2s 44ms/step - loss: 0.2905 - accuracy: 0.8996\n",
      "Epoch 87/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.2600 - accuracy: 0.9197\n",
      "Epoch 88/200\n",
      "42/42 [==============================] - 2s 45ms/step - loss: 0.2373 - accuracy: 0.9241\n",
      "Epoch 89/200\n",
      "42/42 [==============================] - 2s 45ms/step - loss: 0.2184 - accuracy: 0.9289\n",
      "Epoch 90/200\n",
      "42/42 [==============================] - 2s 43ms/step - loss: 0.2027 - accuracy: 0.9329\n",
      "Epoch 91/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.1917 - accuracy: 0.9345\n",
      "Epoch 92/200\n",
      "42/42 [==============================] - 2s 44ms/step - loss: 0.1318 - accuracy: 0.9546\n",
      "Epoch 93/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.1282 - accuracy: 0.9614\n",
      "Epoch 94/200\n",
      "42/42 [==============================] - 2s 43ms/step - loss: 0.1672 - accuracy: 0.9438\n",
      "Epoch 95/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.1654 - accuracy: 0.9502\n",
      "Epoch 96/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.1133 - accuracy: 0.9622\n",
      "Epoch 97/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.1221 - accuracy: 0.9602\n",
      "Epoch 98/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.1172 - accuracy: 0.9683\n",
      "Epoch 99/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.1318 - accuracy: 0.9590\n",
      "Epoch 100/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0934 - accuracy: 0.9711\n",
      "Epoch 101/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0889 - accuracy: 0.9723\n",
      "Epoch 102/200\n",
      "42/42 [==============================] - 2s 43ms/step - loss: 0.0935 - accuracy: 0.9667\n",
      "Epoch 103/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.1061 - accuracy: 0.9671\n",
      "Epoch 104/200\n",
      "42/42 [==============================] - 2s 43ms/step - loss: 0.1274 - accuracy: 0.9618\n",
      "Epoch 105/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0945 - accuracy: 0.9687\n",
      "Epoch 106/200\n",
      "42/42 [==============================] - 2s 47ms/step - loss: 0.0684 - accuracy: 0.9783\n",
      "Epoch 107/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0891 - accuracy: 0.9699\n",
      "Epoch 108/200\n",
      "42/42 [==============================] - 2s 43ms/step - loss: 0.1023 - accuracy: 0.9639\n",
      "Epoch 109/200\n",
      "42/42 [==============================] - 2s 45ms/step - loss: 0.1134 - accuracy: 0.9614\n",
      "Epoch 110/200\n",
      "42/42 [==============================] - 2s 45ms/step - loss: 0.1029 - accuracy: 0.9695\n",
      "Epoch 111/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0836 - accuracy: 0.9723\n",
      "Epoch 112/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0860 - accuracy: 0.9743\n",
      "Epoch 113/200\n",
      "42/42 [==============================] - 2s 45ms/step - loss: 0.0925 - accuracy: 0.9743\n",
      "Epoch 114/200\n",
      "42/42 [==============================] - 2s 51ms/step - loss: 0.0745 - accuracy: 0.9767\n",
      "Epoch 115/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0792 - accuracy: 0.9771\n",
      "Epoch 116/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0777 - accuracy: 0.9747\n",
      "Epoch 117/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.1083 - accuracy: 0.9707\n",
      "Epoch 118/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0770 - accuracy: 0.9767\n",
      "Epoch 119/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0742 - accuracy: 0.9763\n",
      "Epoch 120/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0688 - accuracy: 0.9815\n",
      "Epoch 121/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0445 - accuracy: 0.9855\n",
      "Epoch 122/200\n",
      "42/42 [==============================] - 2s 43ms/step - loss: 0.0675 - accuracy: 0.9831\n",
      "Epoch 123/200\n",
      "42/42 [==============================] - 2s 43ms/step - loss: 0.0799 - accuracy: 0.9731\n",
      "Epoch 124/200\n",
      "42/42 [==============================] - 2s 43ms/step - loss: 0.0895 - accuracy: 0.9703\n",
      "Epoch 125/200\n",
      "42/42 [==============================] - 2s 43ms/step - loss: 0.0768 - accuracy: 0.9759\n",
      "Epoch 126/200\n",
      "42/42 [==============================] - 2s 43ms/step - loss: 0.0530 - accuracy: 0.9839\n",
      "Epoch 127/200\n",
      "42/42 [==============================] - 2s 44ms/step - loss: 0.0477 - accuracy: 0.9863\n",
      "Epoch 128/200\n",
      "42/42 [==============================] - 2s 47ms/step - loss: 0.0590 - accuracy: 0.9819\n",
      "Epoch 129/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0818 - accuracy: 0.9779\n",
      "Epoch 130/200\n",
      "42/42 [==============================] - 2s 43ms/step - loss: 0.0808 - accuracy: 0.9711\n",
      "Epoch 131/200\n",
      "42/42 [==============================] - 2s 43ms/step - loss: 0.0661 - accuracy: 0.9783\n",
      "Epoch 132/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0568 - accuracy: 0.9827\n",
      "Epoch 133/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0504 - accuracy: 0.9835\n",
      "Epoch 134/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0539 - accuracy: 0.9835\n",
      "Epoch 135/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0510 - accuracy: 0.9835\n",
      "Epoch 136/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0408 - accuracy: 0.9827\n",
      "Epoch 137/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0730 - accuracy: 0.9811\n",
      "Epoch 138/200\n",
      "42/42 [==============================] - 2s 44ms/step - loss: 0.0546 - accuracy: 0.9835\n",
      "Epoch 139/200\n",
      "42/42 [==============================] - 2s 45ms/step - loss: 0.0429 - accuracy: 0.9876\n",
      "Epoch 140/200\n",
      "42/42 [==============================] - 2s 47ms/step - loss: 0.0446 - accuracy: 0.9859\n",
      "Epoch 141/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0428 - accuracy: 0.9863\n",
      "Epoch 142/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0584 - accuracy: 0.9827\n",
      "Epoch 143/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0420 - accuracy: 0.9888\n",
      "Epoch 144/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 0.0541 - accuracy: 0.9843\n",
      "Epoch 145/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 0.0431 - accuracy: 0.9851\n",
      "Epoch 146/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0424 - accuracy: 0.9888\n",
      "Epoch 147/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0778 - accuracy: 0.9787\n",
      "Epoch 148/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0815 - accuracy: 0.9787\n",
      "Epoch 149/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0542 - accuracy: 0.9827\n",
      "Epoch 150/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 0.0534 - accuracy: 0.9835\n",
      "Epoch 151/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 0.0362 - accuracy: 0.9855\n",
      "Epoch 152/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 0.0422 - accuracy: 0.9859\n",
      "Epoch 153/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0329 - accuracy: 0.9912\n",
      "Epoch 154/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0367 - accuracy: 0.9892\n",
      "Epoch 155/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0429 - accuracy: 0.9867\n",
      "Epoch 156/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0330 - accuracy: 0.9904\n",
      "Epoch 157/200\n",
      "42/42 [==============================] - 2s 44ms/step - loss: 0.0269 - accuracy: 0.9920\n",
      "Epoch 158/200\n",
      "42/42 [==============================] - 2s 43ms/step - loss: 0.0341 - accuracy: 0.9884\n",
      "Epoch 159/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0366 - accuracy: 0.9896\n",
      "Epoch 160/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 0.0349 - accuracy: 0.9863\n",
      "Epoch 161/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0248 - accuracy: 0.9928\n",
      "Epoch 162/200\n"
     ]
    },
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0395 - accuracy: 0.9871\n",
      "Epoch 163/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0568 - accuracy: 0.9799\n",
      "Epoch 164/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 0.0436 - accuracy: 0.9876\n",
      "Epoch 165/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0404 - accuracy: 0.9880\n",
      "Epoch 166/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0407 - accuracy: 0.9876\n",
      "Epoch 167/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 0.0327 - accuracy: 0.9896\n",
      "Epoch 168/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0355 - accuracy: 0.9863\n",
      "Epoch 169/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 0.0473 - accuracy: 0.9867\n",
      "Epoch 170/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 0.0349 - accuracy: 0.9892\n",
      "Epoch 171/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 0.0421 - accuracy: 0.9888\n",
      "Epoch 172/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0527 - accuracy: 0.9855\n",
      "Epoch 173/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0496 - accuracy: 0.9876\n",
      "Epoch 174/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 0.0455 - accuracy: 0.9847\n",
      "Epoch 175/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 0.0282 - accuracy: 0.9928\n",
      "Epoch 176/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 0.0345 - accuracy: 0.9888\n",
      "Epoch 177/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 0.0268 - accuracy: 0.9904\n",
      "Epoch 178/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0468 - accuracy: 0.9855\n",
      "Epoch 179/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0371 - accuracy: 0.9871\n",
      "Epoch 180/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0537 - accuracy: 0.9835\n",
      "Epoch 181/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0350 - accuracy: 0.9892\n",
      "Epoch 182/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0518 - accuracy: 0.9863\n",
      "Epoch 183/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0350 - accuracy: 0.9884\n",
      "Epoch 184/200\n",
      "42/42 [==============================] - 2s 44ms/step - loss: 0.0324 - accuracy: 0.9904\n",
      "Epoch 185/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0294 - accuracy: 0.9908\n",
      "Epoch 186/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0421 - accuracy: 0.9876\n",
      "Epoch 187/200\n",
      "42/42 [==============================] - 2s 43ms/step - loss: 0.0459 - accuracy: 0.9839\n",
      "Epoch 188/200\n",
      "42/42 [==============================] - 2s 46ms/step - loss: 0.0408 - accuracy: 0.9888\n",
      "Epoch 189/200\n",
      "42/42 [==============================] - 2s 44ms/step - loss: 0.0282 - accuracy: 0.9920\n",
      "Epoch 190/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0355 - accuracy: 0.9884\n",
      "Epoch 191/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0401 - accuracy: 0.9880\n",
      "Epoch 192/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0312 - accuracy: 0.9888\n",
      "Epoch 193/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0287 - accuracy: 0.9920\n",
      "Epoch 194/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0456 - accuracy: 0.9855\n",
      "Epoch 195/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0318 - accuracy: 0.9892\n",
      "Epoch 196/200\n",
      "42/42 [==============================] - 2s 42ms/step - loss: 0.0478 - accuracy: 0.9876\n",
      "Epoch 197/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0385 - accuracy: 0.9908\n",
      "Epoch 198/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0659 - accuracy: 0.9759\n",
      "Epoch 199/200\n",
      "42/42 [==============================] - 2s 40ms/step - loss: 0.0462 - accuracy: 0.9880\n",
      "Epoch 200/200\n",
      "42/42 [==============================] - 2s 41ms/step - loss: 0.0364 - accuracy: 0.9863\n"
     ]
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "source": [
    "plt.plot(history.history[\"loss\"])\n",
    "plt.ylabel('Categorical crossentropy loss')\n",
    "plt.xlabel('epochs')"
   ],
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "Text(0.5, 0, 'epochs')"
      ]
     },
     "metadata": {},
     "execution_count": 27
    },
    {
     "output_type": "display_data",
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEGCAYAAABo25JHAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8rg+JYAAAACXBIWXMAAAsTAAALEwEAmpwYAAAwaElEQVR4nO3deXxU9fX/8dfJZN8hG4SENQEEAYGgoIjiVrQKdde61LbfqrWtxS5qbaut+uuubV3qbtWqdatW6r4hKAqyyI5AWAx7EkJWQrY5vz/mggGyXCAzN8mc5+NxH5m5c+fOe24mc3Lv597PR1QVY4wx4SvC6wDGGGO8ZYXAGGPCnBUCY4wJc1YIjDEmzFkhMMaYMBfpdYBDlZ6erv379/c6hjHGdCkLFy4sVdWMlh7rcoWgf//+LFiwwOsYxhjTpYjIl609ZoeGjDEmzFkhMMaYMGeFwBhjwpwVAmOMCXNWCIwxJsxZITDGmDBnhcAYY8Jc2BSC4so9/O6NVWwtr/U6ijHGdCphUwjmbijjsY83MOlPM/nnnA1exzHGmE4jbArB1FHZzPr5yYzMSeGR2euxAXmMMSYgbAoBQE6PeC4Ym8vWij2sK6n2Oo4xxnQKYVUIACYNTgdg1ppSj5MYY0znEHaFIKdHPAMzEpi1psTrKMYY0ymEXSEAOGlwBvPW72RPQ5PXUYwxxnNhWQgmDc6grtHP3PU7vY5ijDGeC8tCMGFgGvHRPt5ducPrKMYY47mwLASxUT4m5Wfw3qoddhqpMSbshWUhADh9WBY7KutYtqXC6yjGGOOpsC0Ek4dmEiHY4SFjTNgL20LQMyGasf16MNtOIzXGhLmwLQQAY/r2YNW2Kuob/V5HMcYYz4R1IRiRk0J9k5/V26u8jmKMMZ4J60Iwsk8qAEu3lHuawxhjvBTWhSC3ZxwpcVEs22xnDhljwldYFwIRYWROCkutEBhjwljQCoGI5IrITBFZKSIrROTHLSxzsohUiMhiZ7o1WHlaMzInhTU7qqzfIWNM2IoM4robgZ+q6iIRSQIWisi7qrrygOU+UtWzg5ijTSP6pNLoV1ZsrWRsvx5exTDGGM8EbY9AVbep6iLndhWwCugTrNc7XMcO6ElkhNiFZcaYsBWSNgIR6Q+MBua18PAEEVkiIm+KyPBWnn+1iCwQkQUlJR17AVjPhGgm5qfzvyVb8fut3yFjTPgJeiEQkUTgP8B0Va084OFFQD9VHQXcC/y3pXWo6sOqWqCqBRkZGR2ecdox2Wwpr2VR0a4OX7cxxnR2QS0EIhJFoAg8o6ovH/i4qlaqarVz+w0gSkTSg5mpJacP60VMZASvLt4a6pc2xhjPBfOsIQEeA1ap6t2tLNPLWQ4ROdbJE/LRYhJjIjllaCbvrNxu3VIbY8JOMM8aOgG4AlgmIoudebcAfQFU9UHgAuD7ItII1AKXqEffxBPz03lz+XY2lNYwMCPRiwjGGOOJoBUCVf0YkHaWuQ+4L1gZDsWEgWkAfLp+pxUCY0xYCesri5sbkJ5Ar+RYPlln4xgbY8LLIRUCEekhIiODFcZLIsKEQWnMXbfT2gmMMWGl3UIgIh+KSLKI9CRwuucjItJi429XN2FQGjtr6lmzo9rrKMYYEzJu9ghSnPP/zwOeUtXjgNOCG8sbE/PSEYE3lm3zOooxxoSMm0IQKSK9gYuA14Kcx1PZqXGcmJ/B8/M30dhko5YZY8KDm0JwO/A2UKiq80VkILA2uLG8881j+7K9cg8zV9tYxsaY8NBuIVDVF1V1pKpe59xfr6rnBz+aN049KpPMpBienfel11GMMSYk3DQW/8lpLI4SkfdFpERELg9FOC9E+SKYdkw2HxeWsru+0es4xhgTdG4ODZ3hNBafDWwE8oCfBzOU1ybmZ9DQpHy2oczrKMYYE3SuGoudn18HXlTVbj+u47j+PYj2RdjFZcaYsOCmi4nXROQLAn0BfV9EMoA9wY3lrfjoSEb3TWVOYanXUYwxJujcNBbfDBwPFKhqA1ADTAt2MK+dkJfOym2VlNXUex3FGGOCyk1jcRRwOfC8iLwEfBcPuooOtRPy0lHF9gqMMd2emzaCB4CxwD+caYwzr1sblZNCemIMry21wWqMMd2bmzaCcc5Qknt9ICJLghWos4j0RTB1VDb/mruR8t31pMZHex3JGGOCws0eQZOIDNp7x7myuCl4kTqP88b0oaFJeW2p9T1kjOm+3BSCnwMznV5IZwEfAD8NbqzOYXh2MoOzEnl50WavoxhjTNC0e2hIVd8XkXxgiDNrtarWBTdW5yAiTDumD39+ezVby2vJTo3zOpIxxnS4VvcIROS8vROBi8nynOnrzrywMOXoXgC8s2K7x0mMMSY42tojOKeNxxR4uYOzdEqDMhLJz0zkrRXbueqEAV7HMcaYDtdqIVDVb4cySGc25ehe3D+zkJ3VdaQlxngdxxhjOpQNXu/ClKN74Vd4f1Wx11GMMabDWSFwYVjvZNISovlso/VGaozpfqwQuCAijO6byudFu7yOYowxHc5NX0MLReQHItIjFIE6q2NyU1lXUkNFbYPXUYwxpkO52SO4GMgG5ovIcyLyNRGRIOfqdEb3DdTBJZvKvQ1ijDEdzE031IWq+ktgMPAs8DjwpYj8VkR6BjtgZzEyJwUR+Lyo3OsoxhjToVy1EYjISOAu4M/Af4ALgUoC3U209pxcEZkpIitFZIWI/LiFZURE7hGRQhFZKiJjDu9tBF9SbBSDM5P4fJO1Exhjupd2u5gQkYVAOfAYcHOz7iXmicgJbTy1Efipqi4SkSRgoYi8q6ormy1zJpDvTMcR6N76uEN/G6Exum8qby7fjt+vRESE3dExY0w35WaP4EJVPVVVnz2wjyFVbbWrCVXdpqqLnNtVwCqgzwGLTQOe0oC5QKqI9D60txA6xw3sSUVtAyu2VnodxRhjOoybQlDhHL5Z5JxB9HcRSTuUFxGR/sBoYN4BD/UBNjW7v5mDiwUicrWILBCRBSUlJYfy0h1qUn4GIvDharuwzBjTfbgpBM8BJcD5wAXO7efdvoCIJBJoV5iuqof1r7SqPqyqBapakJGRcTir6BBpiTGM6JPCh2u8K0bGGNPR3BSC3qp6h6pucKY7gSw3K3fGO/4P8IyqttRJ3RYgt9n9HGdep3Xy4Aw+L9pFxW67nsAY0z24KQTviMglIhLhTBcBb7f3JOdag8eAVap6dyuLzQCudM4eGg9UqGqnHg7spCEZ+BU+KrS9AmNM9+BmzOLvAdOBp537EUCNiFwDqKomt/K8E4ArgGUistiZdwvQl8ATHwTeAM4CCoHdQKfv8fSY3B4kRPuYv6GMs0dmex3HGGOOmJsRypIOZ8Wq+jHQ5jmWqqrADw5n/V7xRQh5WUmsLa72OooxxnQIN3sEiMhUYJJz90NVfS14kTq//MxEZlmDsTGmm3DT6dwfgB8DK53pxyLy+2AH68wGZyVSUlVH+e56r6MYY8wRc9NYfBZwuqo+rqqPA1MIjGEctvIzA0fL7PCQMaY7cDseQWqz2ylByNGl5GclArB2hxUCY0zX56aN4HfA5yIyk0Dj7yTg5qCm6uSyU+KIj/axZkeV11GMMeaItVkIRCQC8APjgXHO7JtUdXuwg3VmERFCfmYihXZoyBjTDbRZCFTVLyI3quoLBC7+Mo68zCQ+WmtnDhljuj43bQTvicjPnPEFeu6dgp6sk8vLTKS4qo7KPdbVhDGma3PTRnCx87P5hV8KDOz4OF1Hv7R4ADaV7WZ4dti3nxtjujA3heAoVd3TfIaIxAYpT5eR2yNQCDbvqrVCYIzp0twcGvrE5bywktMjDgjsERhjTFfW6h6BiPQiMEhMnIiM5qt+g5KB+BBk69RS46NIjIlk865ar6MYY8wRaevQ0NeAqwiMEdC8G+kqAr2IhjURIadHnO0RGGO6vFYLgao+CTwpIuer6n9CmKnLyO0ZT9FOKwTGmK7NTWPxayLyTaB/8+VV9fZgheoqcnrEMaewFFUlMA6PMcZ0PW4KwatABbAQqAtunK4lt0c8u+ubKKupJy0xxus4xhhzWNwUghxVnRL0JF1Qbs+vTiG1QmCM6apcnT4qIiOCnqQL2ncK6S5rJzDGdF1u9ggmAleJyAYCh4aEwCiTI4OarAvYu0ewqcxOITXGdF1uCsGZQU/RRSXGRJKVHMPHhSVce9JAazA2xnRJ7R4aUtUvgVzgFOf2bjfPCxfXTBrEnMKdzFxd7HUUY4w5LG7GLL4NuAn4hTMrCng6mKG6kism9GNgRgJ3vr6Kxia/13GMMeaQufnP/lxgKlADoKpbgaRghupKonwR/PjUfNaX1LB4U7nXcYwx5pC5KQT1qqoEup5GRBKCG6nrOWlwBiIwp3Cn11GMMeaQuSkEL4jIQ0CqiHwPeA94JLixupbU+GiGZyczZ12p11GMMeaQtXvWkKr+RUROByqBIcCtqvpu0JN1MScMSufxORvYXd9IfLSbk7GMMaZzcNNYnAB8oKo/J7AnECciUUFP1sUcn5dOQ5Myf+Mur6MYY8whcXNoaDYQIyJ9gLeAK4An2nuSiDwuIsUisryVx08WkQoRWexMtx5K8M5mXP8eRPmEj9bYgPbGmK7FTSEQVd0NnAc8oKoXAsNdPO8JoL0+ij5S1WOcqUv3ZhofHckpQzN5fv4mdtXUex3HGGNcc1UIRGQCcBnwujPP196TVHU2UHYE2bqcn54xhOr6Rv7xYaHXUYwxxjU3hWA6gYvJXlHVFSIyEJjZQa8/QUSWiMibItLqXoaIXC0iC0RkQUlJ5z30MjgrifNG5/Dkp1+ys9p67DbGdA1uupiYpapTVfWPIhIBlKrq9R3w2ouAfqo6CrgX+G8bGR5W1QJVLcjIyOiAlw6e88f0ob7Rzxfbq7yOYowxrrg5a+hZEUl2zh5aDqwUkZ8f6QuraqWqVju33wCiRCT9SNfrta96JLWuqY0xXYObQ0PDVLUS+AbwJjCAwJlDR0REeonTXaeIHOtk6fKX5vZOicUXIRRZITDGdBFurnyKcq4b+AZwn6o2iIi29yQR+TdwMpAuIpuB2wh0WIeqPghcAHxfRBqBWuASpyuLLi3SF0F2aiybdtkYBcaYrsFNIXgI2AgsAWaLSD8CVxm3SVUvbefx+4D7XLx+l9O3Z7wdGjLGdBluGovvUdU+qnqWBnwJTA5Bti4rt4cVAmNM1+GmsThFRO7ee/qmiNwFWA+kbcjtGc/Omnpq6hq9jmKMMe1y01j8OFAFXORMlcA/gxmqq9t35pANam+M6QLctBEMUtXzm93/rYgsDlKebqFvs0Hth/ZK9jiNMca0zc0eQa2ITNx7R0ROIHCWj2lFbo84IHAtQTc4EcoY0825KQTXAveLyEYR2UjgTJ9rgpqqi+uZEE18tI+HZq9j6K/forC42utIxhjTqjYPDYmID7hCVUeJSDIErggOSbIuTETIz0xk5bZKGpqUT9eVkpeZ6HUsY4xpUZt7BKraBEx0bldaEXDvwSvGMvvGyaQnRrNkc4XXcYwxplVuGos/F5EZwItAzd6Zqvpy0FJ1A71TAu0EI3NSWbKp3NswxhjTBjeFIJZAH0CnNJungBUCF0bmpDBzdTHVdY0kxthYxsaYzsfN4PXfDkWQ7mpUTiqqsHxLBeMHpnkdxxhjDuLmyuInRSS12f0eIvJ4UFN1IyNzUgBYurnc2yDGGNMKN6ePjlTV8r13VHUXMDpoibqZtMQYcnrEsWSTNRgbYzonN4UgQkR67L0jIj1x17ZgHGP69uCzjWV2cZkxplNyUwjuAj4VkTtE5A7gE+BPwY3VvUzMS6ekqo61dmGZMaYTctMN9VPAecAOZzpPVf8V7GDdyQn5gRE4P15b6nESY4w5mKtDPKq6ElgZ5CzdVp/UOPqnxTOnsJTvTBzgdRxjjNmPm0NDpgOckJfO3PU7aWjyex3FGGP2Y4UgRCbmpVNT38SE33/AL19Z5nUcY4zZxwpBiEwemsl1Jw9iYEYCz8wrsqEsjTGdRquFQESqRKSyhalKRKzzuUMUG+XjxilD+evFxxAh8Pz8TV5HMsYYoI1CoKpJqprcwpSkqjbs1mHqkxrHSYMzeHHhJhqtvcAY0wm4PjQkIpki0nfvFMxQ3d2lx/ZlR2Udry/b5nUUY4xx1dfQVBFZC2wAZgEbgTeDnKtbO2VoJsOzk/ndG6uo2tPgdRxjTJhzs0dwBzAeWKOqA4BTgblBTdXNRfoi+N25IyiuquPPb6/2Oo4xJsy5KQQNqrqTQJ9DEao6EygIcq5ub1RuKlcd35+nPv2SN+0QkTHGQ26uLC4XkURgNvCMiBTTbKQyc/huPnMonxeV87MXlzC4VxKDMmxcY2NM6LnZI5gG7AZuAN4C1gHntPckEXlcRIpFZHkrj4uI3CMihSKyVETGHErw7iAm0scDl4/BFyH8ZsYK653UGOMJN4UgE4hW1UZVfRJ4BEhy8bwngCltPH4mkO9MVwMPuFhnt9M7JY7rT83no7WlfLimxOs4xpgw5KYQvAg0P+G9yZnXJlWdDZS1scg04CkNmAukikhvF3m6nSsn9GdAegK3vbqCop12xbExJrTcFIJIVa3fe8e5Hd0Br90HaH557WZn3kFE5GoRWSAiC0pKut9/zdGREfzlwpFU1DZwzn0fs6hol9eRjDFhxE0hKBGRqXvviMg0IKQd66vqw6paoKoFGRkZoXzpkBnbryf/++FEUuKi+P7TC9lZXed1JGNMmHBTCK4FbhGRIhHZBNwEXNMBr70FyG12P8eZF7b6psXzwOVj2LW7genPL6bJb43HxpjgczNC2TpVHQ8MA45S1eNVtbADXnsGcKVz9tB4oEJVw/6E+uHZKfx26nA+WlvKHa/ZWEDGmOBr9ToCEblcVZ8WkZ8cMB8AVb27rRWLyL+Bk4F0EdkM3AZEOc99EHgDOAsoJHB66rcP+110M5ce25fC4moe+3gDPeKjuf7UvH3b3RhjOlpbF5QlOD/dnCp6EFW9tJ3HFfjB4aw7HNxy1lHsqqnnr++tYUNpNbd8/Sgyk2K9jmWM6YZaLQSq+pCI+IBKVf1rCDMZwBch3HXRKPqmxXPP+2t5c/l2fnfuCM4fm+N1NGNMN9NmG4GqNgFt/mdvgkdEmH7aYD746cmM6JPCbTNWsKNyj9exjDHdjJuzhuaIyH0icqKIjNk7BT2Z2ad/egJ/uXAU9U1+/t/rq7yOY4zpZtx0OneM8/P2ZvMUOKXD05hW9U9P4LqTB/G399YyPDuZa04aBEB9ox8RiPLZ8NPGmMPTbiFQ1cmhCGLa98PJeawrqeH3b36BApeMy+Xcf3zCoIxEHv2W9QxujDk87RYCEUkhcOrnJGfWLOB2Va0IZjBzsEhfBH+9aBQC/OHNL3h2XhFFZbvZUFrDss0VjMhJ8TqiMaYLcnM84XGgCrjImSqBfwYzlGldpC+Cuy8axdRR2RSV7eaXZx1FUkwk988s5K3l21n4pfVTZIw5NNJeH/gislhVj2lvXqgUFBToggULvHjpTqXJr2zcWcOgjET++NYXPPDhOgASYyJ5a/qJ5PSI37dsbX0Tu+sbSUuM8SquMcZjIrJQVVs8huxmj6BWRCY2W9kJQG1HhTOHxxch+0Y0u2bSQP5v4gD+evEoVJWfv7iUPQ1N+5a9+eWlnPHX2daRnTGmRW7OGvo+8KTTViAExhi4KpihzKFJjY/mV2cPA6Cuwc/NLy9j1G/f4coJ/fjB5DzeXLad+iY/v/nfSu69dLTHaY0xnY2bs4YWA6NEJNm5XxnsUObwXTwulz494nh2XhGPfLSBHZV11Df5OWtEL/63ZCsT89K4eFzffcs3NPl54MN1nD4si6N6J3uY3BjjFTdnDbXU6VwFsNApEqYTERFOzM9gXP+eLN08ixlLtjIkK4m/XTyaytr53PSfZdTUNfGdiQPw+5UbX1rKK59vYfmWCh6+0k5BNSYcuWkjKCAwJkEfZ7qGwFjEj4jIjUHMZo5AbJSPG6cMAeCCsTlER0bw2FUFnHl0L25/bSUPzVrH9c99ziufb6FPahyz15ZQW9900Hp21zcyp7CU9k4qMMZ0XW7aCHKAMapaDSAitwGvE7iuYCHwp+DFM0di6qhskuOiOH5QGgAxkT7uuXQ01z2ziN+/+QURAjdOGcKIPilc8dhnfFxYys7qOmatKaFyTwPjB6TxyudbWF9aw/TT8pl+2mCP35ExJhjcFIJMoPnpJg1AlqrWioidhtKJiQiTh2TuNy/KF8F93xzNve8XMjE/nfED02ho8pMUG8ntr61gU1ktuT3jSIiO5K5319A7JZZTh2byt/fWkhQbxZUT+hHli6C2vok3l2/jrBG9iY3yefQOjTEdwU0heAaYJyKvOvfPAZ4VkQTAhtDqgmIiffzsa0P23Y/yRTB5SCYzlmxl8pAMHv3WOHwRQnHlHpJio/BFCN99cj53vLaSxz5azwVjc3hn5Q6+2F7F8i2V3HrOMA/fjTHmSLV7QRmAiBQAJzh356iqZ1d02QVlwbF8SwVPfrKRW88ZRlJs1EGPqyofrinh0Y/W88m6nSTHRjEyJ4U5haW8eO3xDMpIICEmcl/nd7vrGymrqd/vwjZjjHfauqDMbSGYCOSr6j9FJANIVNUNHZzTFSsE3iuu3ENstA9VOO3uWZRUfXWEMC8zkUvG5fLEJxvZWl7LlRP6c+OUIcRHt77zufDLXWQkxtA3zYqGMcHSViFwc/robQTOHBpCoI+hKOBpvtpDMGEmM/mrITP//b3jeG9VMVG+CKr2NPDmsu3c+foq+vaM54KxOTz56UYa/X7u/MaI/dZRWFxNz4RoSqvruPThuWSlxPDO9JOIi7b2BmNCzU0bwbnAaGARgKpuFZHDGsfYdD95mUnkZX71cfjRKfnMW7+TkbmpJMZEEhPp49+fFXFRQS6PfLSBM4ZlMTw7ma/f8xFJsVGkJUQTExnBprJa/vb+Gn5x5lEevhtjwpObQlCvqioiCuA0EhvTIl+EcHxe+r77PzoljxcXbuLcf3xCk195a/k2+qclEBMZQXJsJKt3VHHvpaP5eG0pj360gdOPyqKgf08P34Ex4cfNBWUviMhDQKqIfA94D3g0uLFMd5GZHMvVkwYRH+XjwcvH0rdnPGuLq7n1nOHM+NFEnrt6PGeP7M0tXz+K3B5xXPv0Ih6ZvZ7/e3IBRTt3ex3fmLDgtrH4dOAMAp3Ova2q7wY7WGussbjrUVXqGv3ERvnYVlHLvPVlTDsme293Jfus2VHFuffPoaa+yeldNYEXrz0eVSU1PhpV5dN1OxnbvwcxkdaWYMyhOKKzhkTkj6p6U3vzQsUKQff2xfZKGhqVitoGvvXPz2jyBz6f1540iMYmP49+vIHzx+Rw10WjPE5qTNdyRGcNAacDB37pn9nCPGOO2NBeX/WA+siVY/lswy62V9Ty4KzAwDtDspL4z6LNHD8ojfPH5ngV05hupdVCICLfB64DBorI0mYPJQFzgh3MmFOGZnHK0CxUlaP7pFC+u4Efn5bPZY/O46b/LGV3fSOXj+930CEmY8yhafXQkDMQTQ/g98DNzR6qUtWyEGRrkR0aMhW1DUx/7nNmri7hknG5/HbacGszMKYdR3xlsbOSTGDflUSqWuTiOVOAvwM+4FFV/cMBj18F/BnY4sy6T1XbPCPJCoGBwJjNd7+7mvtnruOo3slcMb4fW8trqa5r5NdnD8MXYXsJxjR3pFcWnwPcDWQDxUA/YBUwvJ3n+YD7CbQxbAbmi8gMVT2wo7rnVfWH7b4LY5rxRQg//9pQjs5O4c9vr+aWV5YhAqqQn5XIZcf18zqiMV2Gm8biO4HxwHuqOlpEJgOXu3jesUChqq4HEJHngGlYj6WmA505ojdTju7Fym2V9E6J47pnFvKnt1YzZXgv0hJjvI5nTJfg5oKyBlXdCUSISISqziTQ91B7+gCbmt3f7Mw70PkislREXhKR3JZWJCJXi8gCEVlQUlLi4qVNOBERhmen0DMhmjumHU1NXSN/e2+t17GM6TLcFIJyEUkEZgPPiMjfgZoOev3/Af1VdSTwLvBkSwup6sOqWqCqBRkZGR300qY7ys9K4uJxuTw3v4hNZXZlsjFuuCkE04DdwA3AW8A6AoPTtGcL0Pw//By+ahQGQFV3qurePowfBca6WK8xbfrhKXmICPd+YHsFxrjRaiEQkTwROUFVa1TVr6qNqvokgV5IU12sez6QLyIDRCQauASYccBr9G52dyqBRmhjjkjvlDguO64vLy3czKKiXV7HMabTa2uP4G9AZQvzK5zH2qSqjcAPgbcJfMG/oKorROR2EZnqLHa9iKwQkSXA9cBV7qMb07rppw2md0ocNzy/mOq6Rq/jGNOptXVB2XxVHdfKY8tUdURLjwWbXUdg3PpsQxmXPPwppwzN5P7LxthFZyastXUdQVt7BKltPBZ3RImMCYFjB/Tkt1OH896qYr7/9CL8fncXTxoTbtoqBAuc8Qf2IyL/BywMXiRjOs4VE/rzizOH8sEXxSzZXO51HGM6pbYKwXTg2yLyoYjc5UyzgO8CPw5JOmM6wHljAr2UztvgWRdZxnRqrRYCVd2hqscDvwU2OtNvVXWCqm4PTTxjjlxGUgwDMxKYt36n11GM6ZTa7WLCuZJ4ZgiyGBM0xw1I47UlW2nyq3VIZ8wB3FxQZkyXN35gT6rqGlm1raUzoo0Jb1YITFg4dkBPAOba4SFjDmKFwISF3ilx5GUm8uCs9Xyx3fYKjGnOCoEJGw9cNobICOHih+aysbSj+k00puuzQmDCRn5WEi9cMwFV5YYXFtPY5Pc6kjGdghUCE1b6psVz57kj+LyonL+/b72TGgPuRigzpluZOiqbj9aUcO8HhWQkxXDlhP5eRzLGU1YITFj63XkjKK9t4NZXV9DYpHxn4gCvIxnjGTs0ZMJSlC+Cey8dzZThvbj9tZX8ZsYK6hutzcCEJysEJmzFRvm4/7IxfOeEATzxyUYuePATllrHdCYMWSEwYc0XIdx6zjAevHwsm3fVMvW+Odzx2kqvYxkTUlYIjAGmHN2L2TdOZuqobJ74ZCOl1XX7Hmto8ttYBqZbs0JgjCMxJpIfTM6jya+8vnQbAJt37WbM7e8y+FdvcvFDn1pBMN2SFQJjmhnSK4mhvZL47+ItAPzl7dXUN/mZOiqbeRvK+Liw1OOExnQ8KwTGHOAbo/vweVE5976/lv8u3sp3Jw7g9+ePIC0hmqfnfrnfsoXF1by0cDOtjf1dXddIbX1TKGIbc9jsOgJjDnDu6D78c84G7np3DT0Torn25EHERPq4sCCXh2ev4xcvL2NdcTX90uKZsWQrdY1+kmMjOWN4r/3WU1hczaWPzKV3Siwvf/94In0H/99VXLWH9IQYImyMBOMhae0/mc6qoKBAFyxY4HUM0801NvkpKttNXLSP3ilxABTt3M3Jf5lJZEQEQ3snsXZHNRPz01lfUo0qvH3DJKKcL/tP1pVy/b8Xs6ehieq6Rm4+cyjXnjRo3/pXbK3gj2+tZvaaEgZnJfLzrw3l9GFZ7Wa65ZVlHDcgjfPH5gTvzZtuSUQWqmpBi49ZITDGvZVbK8lKjiEtMQZVRUR4f9UOvvvkAk4Zmsmw3sks3lTOx4Wl5PSI44lvj+NPb61m1poS/vejifRLi+dPb63mn3M2kBIXxcXj+vLeqh0UFldz05RAMVi2pZzPNuzirBG9ODE/Y99r3/v+Wu56dw3Rvgheu34ig7OSWs25raKW2WtKmDw0k8yk2FBsGtPJWSEwJohUld+9sYpXF2+luKqOvMxEpo7K5upJA4mN8lFcuYev3/sx8dE++qUlMHtNCZcd15cbpwwlJS6KusYmfvbiUv63ZOu+dUb5hCa/8s3j+rKjsg6/X5m1poSTh2SwqKic9MRoLirIZXh2CqP7phIb5QOgpKqOu95ZzUsLN9PoV5JiIvnV2Udx8bi+++X923treWZeEQkxPi4qyOWaSQNbPHS1l9+vrNpeybDeyYgc2mEsv1/5yzurmVNYSlJsFPd/cwwp8VGHuJXNkbJCYEyI1DU2ERPpO2j+wi93cenDc2nw+/n9uSO45Ni++z3u9yuvL9uGX5WB6YkMyEjgpy8s5u0VOxiYkUBspI+0xGjuu3QMC4vK+MkLSyjf3QBAUkwkj397HI1NytVPLWBPYxOXHdePKUf34p731/LJup3cdeEozh+bg6ryjw/X8ee3V3Nifjqq8HFhKSP6pHDdyYM4bVjWvsNbzd397hrueX8tFxfkcue5Rx+0TPnueuoa/WQlH7z38e/PivjFy8so6NeDJZvLOTE/g0evLOjQdhG/X9nd0ERiTPdp9qypa+SWV5Zx+rAszh6ZfcTrs0JgTCcwp7CUhiY/Jw/JdP2c1gqLqlJWU8+SzeXc+doqdtbU0+RXeqfE8tAVYxmYkbjv+d95Yj7z1pfxteG92FZRy6Kicr5xTDZ3X3QMERHCjCVb+fPbX7CprJbk2EiGZSezqayWmMgIBmclcdKQDH793+X0S4tnXUkNI3NS+OkZQ/Z96ZZU7eGXryyntqGJm6YM5ZN1pWyvrON7Jw5gcFYSFz30KYOzknj+6vE8PfdLfv3qClLioujbM54/nj+So3onUdfo37dXA4EvwUiftPjeD1RctYfrnl7Eiq2VTD8tnwsLcumZEH3QclvKa1lcVM6pR2USG+VjyaZyHvhwHWU19eRlJTI4M5GThmQyID1hv+dtKK1hQ2k1xw9K549vfcHyLRX87ZLRzCks5eO1pfzq60eReUAB3FZRyxfbq0iOjWREn1SiI1vf21JVPttQxsicVHwRwuvLtpKXkcQjH61nxpKtiMAvzzqKq47v3+ZeW3usEBjTjRXt3M15D8whNsrHS9ceT6+U/b+UqvY08P9eX8WsNSVEiPDDU/K4qCAXX7P/yJv8yswvinl7xXbWOmdE1Tf6mb9xF6XVdWQkxfDuDZP4aG0pd76+kh2Vdfu9xtBeSSTGRLLgy10kxkSSmRTDemcUOF+E8Mb1JzKkVxKqyrOfFfHFtireWbmditoG0hNj2FaxhylH9yInNY7Fm8pZ+OUukmIjuWJCf84Z2Zu6Rj/vryqmqGw39U1+0hOjmToqm9r6Jn7ywhIqahsY0y+VOYWBMaljIiNQYFROCicNzmBLeS0vL9pCXaOfPqlxpMRFsXJbJanxUeRlJLJmRxWVexqJjYrg9qlH41elwa/k9Ihj+nOLqahtIC7KR21DE7FREUSIsNs5LTgjKYYrx/ejb1o8GYkx/POTjby7cse+bZMQ7eO6yXnkZSby7sodfF60i8TYKC4dl8uw7GQemrWe15dto19aPClxUSzdXLHvudefms/KrZW8t2oHuT3j+NkZQ5h2TJ/D+px4VghEZArwd8AHPKqqfzjg8RjgKWAssBO4WFU3trVOKwTGHGxndR1RkREkx3bssfc9DU38b8lWhvRKYmROKgC19U3MWlNMjPMf/J76JiYPzcQXIby5fDsTBqbRMyGa2WtLKKmqIz8zkdF9exy07uKqPdz26goampTs1Fj++/kW9jT4GdIriRPy0iksrua9VV99oYpAr+RYYiIj2FFZR21D4It4YHoC935zNMOzU/hsQxlLN5dTXFUXKG6ri1lfUkNiTCRnDMvijOFZPDx7PUpgXIoLC3JJjIlEVdlUVstPXljMgi937ZezX1o8P5icx9vLt3P+2BwGZSRyw/OLOWVoJmeN6M1PX1zCqm1fjYMdF+XjmpMGcvygdMpq6nl50WbecQpDanwUBf16UFS2mzU7qgGIEPjuxAG8s3IHZdX13PGNoymtrqOitoEbThsMwDsrd/DQ7HWcNyaHK8b3O6zfpSeFQER8wBrgdGAzMB+4VFVXNlvmOmCkql4rIpcA56rqxW2t1wqBMd1TQ5Mfgf0Of2wtr3X2ZOCMYb3o4Rzyqalr5Ln5m6itb+S7EwcSF93yISRVpbqukcSYSFeN3PWNft5ftYNBmYFDa7NWlzBtdHa7Z17taWhi867dbNpVy7DeyQe1lSz8sozaej/HDexJlC8CVWXltkqKdu4mt2c8R/dJob7Rz57GplaLuaqiymG3rXhVCCYAv1HVrzn3fwGgqr9vtszbzjKfikgksB3I0DZCWSEwxphD11YhCGYXE32ATc3ub3bmtbiMqjYCFUDagSsSkatFZIGILCgpKQlSXGOMCU9doq8hVX1YVQtUtSAjI6P9JxhjjHEtmIVgC5Db7H6OM6/FZZxDQykEGo2NMcaESDALwXwgX0QGiEg0cAkw44BlZgDfcm5fAHzQVvuAMcaYjhe0y/BUtVFEfgi8TeD00cdVdYWI3A4sUNUZwGPAv0SkECgjUCyMMcaEUFCvx1bVN4A3Dph3a7Pbe4ALg5nBGGNM27pEY7ExxpjgsUJgjDFhrsv1NSQiJcCX7S7YsnSgsw4621mzWa5D01lzQefNZrkOzeHm6qeqLZ5/3+UKwZEQkQWtXVnntc6azXIdms6aCzpvNst1aIKRyw4NGWNMmLNCYIwxYS7cCsHDXgdoQ2fNZrkOTWfNBZ03m+U6NB2eK6zaCIwxxhws3PYIjDHGHMAKgTHGhLmwKQQiMkVEVotIoYjc7GGOXBGZKSIrRWSFiPzYmf8bEdkiIoud6SwPsm0UkWXO6y9w5vUUkXdFZK3z8+AxB4Ofa0iz7bJYRCpFZLoX20xEHheRYhFZ3mxei9tIAu5xPnNLRWRMiHP9WUS+cF77FRFJdeb3F5HaZtvtwRDnavX3JiK/cLbXahH5WrBytZHt+Wa5NorIYmd+KLdZa98RwfucBYY/694TgU7v1gEDgWhgCTDMoyy9gTHO7SQCw3kOA34D/Mzj7bQRSD9g3p+Am53bNwN/7AS/y+1APy+2GTAJGAMsb28bAWcBbwICjAfmhTjXGUCkc/uPzXL1b76cB9urxd+b83ewBIgBBjh/s75QZjvg8buAWz3YZq19RwTtcxYuewTHAoWqul5V64HngGleBFHVbaq6yLldBazi4JHbOpNpwJPO7SeBb3gXBYBTgXWqerhXlx8RVZ1NoKfc5lrbRtOApzRgLpAqIr1DlUtV39HAyH8AcwmMCRJSrWyv1kwDnlPVOlXdABQS+NsNeTYREeAi4N/Bev3WtPEdEbTPWbgUAjfDZoaciPQHRgPznFk/dHbtHvfiEAygwDsislBErnbmZanqNuf2diDLg1zNXcL+f5xebzNofRt1ps/ddwj817jXABH5XERmiciJHuRp6ffWmbbXicAOVV3bbF7It9kB3xFB+5yFSyHodEQkEfgPMF1VK4EHgEHAMcA2AruloTZRVccAZwI/EJFJzR/UwH6oZ+cbS2CAo6nAi86szrDN9uP1NmqJiPwSaASecWZtA/qq6mjgJ8CzIpIcwkid7vfWgkvZ/x+OkG+zFr4j9unoz1m4FAI3w2aGjIhEEfgFP6OqLwOo6g5VbVJVP/AIQdwlbo2qbnF+FgOvOBl27N3NdH4WhzpXM2cCi1R1B3SObeZobRt5/rkTkauAs4HLnC8PnEMvO53bCwkcix8cqkxt/N48316wb9jc84Dn984L9TZr6TuCIH7OwqUQuBk2MyScY4+PAatU9e5m85sf0zsXWH7gc4OcK0FEkvbeJtDQuJz9hxP9FvBqKHMdYL//0rzeZs20to1mAFc6Z3WMByqa7doHnYhMAW4Epqrq7mbzM0TE59weCOQD60OYq7Xf2wzgEhGJEZEBTq7PQpWrmdOAL1R1894ZodxmrX1HEMzPWShawTvDRKBlfQ2BSv5LD3NMJLBLtxRY7ExnAf8CljnzZwC9Q5xrIIEzNpYAK/ZuIyANeB9YC7wH9PRouyUAO4GUZvNCvs0IFKJtQAOBY7HfbW0bETiL437nM7cMKAhxrkICx473fs4edJY93/kdLwYWAeeEOFervzfgl872Wg2cGerfpTP/CeDaA5YN5TZr7TsiaJ8z62LCGGPCXLgcGjLGGNMKKwTGGBPmrBAYY0yYs0JgjDFhzgqBMcaEOSsExgSZiJwsIq95ncOY1lghMMaYMGeFwBiHiFwuIp85/c0/JCI+EakWkb86/cK/LyIZzrLHiMhc+aqv/719w+eJyHsiskREFonIIGf1iSLykgTGB3jGuXoUEfmD0+/8UhH5i0dv3YQ5KwTGACJyFHAxcIKqHgM0AZcRuKJ5gaoOB2YBtzlPeQq4SVVHEriac+/8Z4D7VXUUcDyBK1ch0IPkdAL9yg8EThCRNAJdLAx31nNnMN+jMa2xQmBMwKnAWGC+BEalOpXAF7afrzofexqYKCIpQKqqznLmPwlMcvpq6qOqrwCo6h79qo+fz1R1swY6WltMYKCTCmAP8JiInAfs6w/ImFCyQmBMgABPquoxzjREVX/TwnKH2ydLXbPbTQRGDmsk0PPmSwR6CH3rMNdtzBGxQmBMwPvABSKSCfvGh+1H4G/kAmeZbwIfq2oFsKvZ4CRXALM0MJrUZhH5hrOOGBGJb+0Fnf7mU1T1DeAGYFQQ3pcx7Yr0OoAxnYGqrhSRXxEYoS2CQI+UPwBqgGOdx4oJtCNAoBvgB50v+vXAt535VwAPicjtzjoubONlk4BXRSSWwB7JTzr4bRnjivU+akwbRKRaVRO9zmFMMNmhIWOMCXO2R2CMMWHO9giMMSbMWSEwxpgwZ4XAGGPCnBUCY4wJc1YIjDEmzP1/NIczvtQWdDwAAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     }
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "markdown",
   "source": [
    "## LSTM model evaluation\n",
    "First let us transform the categorical output of the LSTM model in the form of one hot encoding classes to the actual classes."
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "source": [
    "y_pred = lstm_categorical.predict(lstm_x_test) # Gets the LSTM output for the test set\n",
    "y_pred = np.argmax(y_pred, axis=1) # Pass the one hot encoding matrix to actual classes\n",
    "y_test = np.argmax(lstm_y_test, axis=1) # Do the same to the LSTM labels"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "source": [
    "ser = 1 - accuracy_score(y_test, y_pred)\n",
    "print(f'LSTM:\\n SER:\\t {ser:.3f}')\n",
    "acc = accuracy_score(y_test, y_pred)\n",
    "print(f' Acc:\\t {acc:.3f}')"
   ],
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "LSTM:\n",
      " SER:\t 0.023\n",
      " Acc:\t 0.977\n"
     ]
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "source": [
    "conf_mtx = confusion_matrix(y_test, y_pred)\n",
    "\n",
    "conf_df = pd.DataFrame(conf_mtx, index = [i for i in range(M)],\n",
    "                  columns = [i for i in range(M)])\n",
    "plt.figure(figsize = (10,7))\n",
    "sns.heatmap(conf_df, cmap=sns.cm.rocket_r, square=True, linewidths=0.1, annot=True, fmt='d', annot_kws={\"fontsize\": 8})  \n",
    "plt.show()"
   ],
   "outputs": [
    {
     "output_type": "display_data",
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 720x504 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     }
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Now let us re-test the Decision tree with the same inputs as the LSTM model"
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "source": [
    "size = feature_mtx.shape[0]\n",
    "x_train = feature_mtx[:int(0.5*size)] # Split in 50% the dataset in training and test\n",
    "x_test  = feature_mtx[int(0.5*size):]\n",
    "\n",
    "y_train = label_mtx[:int(0.5*size)]\n",
    "y_test  = label_mtx[int(0.5*size):]"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "source": [
    "class_DT = DecisionTreeClassifier(max_depth=100, random_state=42)\n",
    "class_DT.fit(x_train, y_train)"
   ],
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "DecisionTreeClassifier(max_depth=100, random_state=42)"
      ]
     },
     "metadata": {},
     "execution_count": 32
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "source": [
    "y_pred = class_DT.predict(x_test)\n",
    "# y_pred = np.argmax(y_pred, axis=1)\n",
    "# y_test = np.argmax(y_test, axis=1)\n",
    "\n",
    "conf_mtx = confusion_matrix(y_test, y_pred)\n",
    "\n",
    "conf_df = pd.DataFrame(conf_mtx, index = [i for i in range(M)],\n",
    "                  columns = [i for i in range(M)])\n",
    "plt.figure(figsize = (10,7))\n",
    "sns.heatmap(conf_df, cmap=sns.cm.rocket_r, square=True, linewidths=0.1, annot=True, fmt='d', annot_kws={\"fontsize\": 8})  \n",
    "plt.show()"
   ],
   "outputs": [
    {
     "output_type": "display_data",
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 720x504 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     }
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "source": [
    "ser = 1 - accuracy_score(y_test, y_pred)\n",
    "print(f'Decision Tree:\\n SER:\\t {ser:.3f}')\n",
    "\n",
    "acc = accuracy_score(y_test, y_pred)\n",
    "print(f' Acc:\\t {acc:.3f}')"
   ],
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "Decision Tree:\n",
      " SER:\t 0.267\n",
      " Acc:\t 0.733\n"
     ]
    }
   ],
   "metadata": {}
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Conclusion\n",
    "With the values of the accuracy symbol error rate, and the confusion matrix is safe to assume that the LSTM model could understand the cyclic nature of the channel and predict the correct classes for the absolute majority of the test database. Although, the LSTM model is computational costly if compared to simpler models as a decision tree. Thus, the problem should be well analyzed before considering this type of solution, great tools also require great computational power."
   ],
   "metadata": {}
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}