diff --git a/radius_estimation/scripts/csvmerge.py b/radius_estimation/scripts/csvmerge.py
new file mode 100644
index 0000000..8db7068
--- /dev/null
+++ b/radius_estimation/scripts/csvmerge.py
@@ -0,0 +1,21 @@
+import os
+import csv
+
+directory = 'denemedataset2'
+csv_file = 'combined2.csv'
+
+header = ['Radius', 'Depth', 'BBox_X', 'BBox_Y']
+
+# open the combined csv file for writing and write the header row
+with open(csv_file, 'w', newline='') as file:
+    writer = csv.writer(file)
+    writer.writerow(header)
+
+    # loop over each txt file in the directory and append its data to the csv file
+    for filename in os.listdir(directory):
+        if filename.endswith('.txt'):
+            with open(os.path.join(directory, filename), 'r') as txt_file:
+                data = txt_file.read().splitlines()
+                writer.writerows([row.split() for row in data])
+
+print('Combined csv file saved successfully!')
diff --git a/radius_estimation/scripts/dataset_generator.py b/radius_estimation/scripts/dataset_generator.py
index 0677290..bc3283a 100755
--- a/radius_estimation/scripts/dataset_generator.py
+++ b/radius_estimation/scripts/dataset_generator.py
@@ -16,12 +16,12 @@ def __init__(self):
         self.previous_depth = None
         self.image_sub = rospy.Subscriber('/yolov7/yolov7', Detection2DArray, self.detection_callback)
         self.depth_image_sub = rospy.Subscriber('/camera/aligned_depth_to_color/image_raw', Image, self.depth_callback)
-        self.dataset_path = "/home/cenk/datasetxxxxxx/"
+        self.dataset_path = "/home/cenk/denemedataset2/"
         if not os.path.exists(self.dataset_path):
             os.makedirs(self.dataset_path)
     
     def detection_callback(self, detection_array):
-        radius = 6.685
+        radius = 3.1
         for detection in detection_array.detections:
             if detection.results[0].id == 32: # nesne ID'si 33 ise
                 print("detected")
diff --git a/radius_estimation/scripts/predict.py b/radius_estimation/scripts/predict.py
new file mode 100644
index 0000000..0d3fe4f
--- /dev/null
+++ b/radius_estimation/scripts/predict.py
@@ -0,0 +1,12 @@
+# Modeli kaydetme
+import joblib
+# Kaydedilen modeli yükleme
+loaded_model = joblib.load("regression_model.joblib")
+
+# Gerçek zamanlı kullanım
+distance = 0.366
+bbox_x = 205
+bbox_y = 170
+features = [[distance, bbox_x, bbox_y]]
+radius = loaded_model.predict(features)
+print("Tahmin edilen yarıçap:", radius)
diff --git a/radius_estimation/scripts/trainmodel.py b/radius_estimation/scripts/trainmodel.py
new file mode 100644
index 0000000..14373fc
--- /dev/null
+++ b/radius_estimation/scripts/trainmodel.py
@@ -0,0 +1,53 @@
+# Gerekli kütüphaneleri yükleme
+import numpy as np
+import pandas as pd
+from sklearn.model_selection import train_test_split, GridSearchCV
+from sklearn.linear_model import LinearRegression, Ridge, Lasso, ElasticNet
+
+# Verileri yükleme
+data = pd.read_csv("combined.csv", header=None, names=["radius", "depth", "bbox_x", "bbox_y"])
+
+# Verileri özellik ve hedef değişkenleri olarak ayırma
+X = data.drop("radius", axis=1)
+y = data["radius"]
+
+# Train ve test verilerini ayırma
+X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
+
+# Ridge Regresyon modelini eğitme ve optimum hiperparametreleri arama
+ridge = Ridge()
+params = {'alpha': [0.001, 0.01, 0.1, 1, 10]}
+grid_ridge = GridSearchCV(estimator=ridge, param_grid=params, scoring='neg_mean_squared_error', cv=5)
+grid_ridge.fit(X_train, y_train)
+print("Ridge Regresyon Modeli En İyi Parametreler:", grid_ridge.best_params_)
+
+# Lasso Regresyon modelini eğitme ve optimum hiperparametreleri arama
+lasso = Lasso()
+params = {'alpha': [0.001, 0.01, 0.1, 1, 10]}
+grid_lasso = GridSearchCV(estimator=lasso, param_grid=params, scoring='neg_mean_squared_error', cv=5)
+grid_lasso.fit(X_train, y_train)
+print("Lasso Regresyon Modeli En İyi Parametreler:", grid_lasso.best_params_)
+
+# Elastic Net Regresyon modelini eğitme ve optimum hiperparametreleri arama
+elastic = ElasticNet()
+params = {'alpha': [0.001, 0.01, 0.1, 1, 10], 'l1_ratio': [0.1, 0.3, 0.5, 0.7, 0.9]}
+grid_elastic = GridSearchCV(estimator=elastic, param_grid=params, scoring='neg_mean_squared_error', cv=5)
+grid_elastic.fit(X_train, y_train)
+print("Elastic Net Regresyon Modeli En İyi Parametreler:", grid_elastic.best_params_)
+
+# En iyi modeli seçme ve eğitme
+best_model = None
+best_score = -np.inf
+
+for model in [LinearRegression(), grid_ridge.best_estimator_, grid_lasso.best_estimator_, grid_elastic.best_estimator_]:
+    model.fit(X_train, y_train)
+    score = model.score(X_test, y_test)
+    print("Model:", model.__class__.__name__, "R^2 Score:", score)
+    if score > best_score:
+        best_score = score
+        best_model = model
+
+print(best_model)
+# Modeli kaydetme
+import joblib
+joblib.dump(best_model, "regression_model.joblib")