ENH updated product_reviews scripts

scripts/product_reviews/train_topic_models/make_html_viz_for_best_snapshots.sh

@@ -0,0 +1,30 @@
+dataset_subpath="multi_domain_product_reviews_dataset/clean_data_v20180403/"
+dataset_path="$HOME/git/$dataset_subpath/"
+echo "dataset_path:"
+echo "$dataset_path"
+
+export XHOST_SSH_ADDR=mchughes@browncs
+export XHOST_REMOTE_PATH="/nbu/liv/mhughes/slda_results/$dataset_subpath/"
+export XHOST_LOCAL_PATH="/results/$dataset_subpath/"
+
+pushd $PC_REPO_DIR/pc_toolbox/utils_vizhtml/
+
+template_html='<pre>TRAIN AUC=$TRAIN_Y_ROC_AUC<br />VALID AUC=$VALID_Y_ROC_AUC<br /> TEST AUC=$TEST_Y_ROC_AUC<nbsp;></pre>'
+
+for rank_words_by in 'proba_word_given_topic' 'proba_topic_given_word'
+do
+python make_html_collection_from_csv.py \
+    --snapshot_csv_path $XHOST_LOCAL_PATH/best_runs_20180301_pcslda_tensorflow/best_snapshots_PC_sLDA.csv \
+    --html_output_path /tmp/product_reviews_html/rank_words_by="$rank_words_by"/ \
+    --field_order LEGEND_NAME,LABEL_NAME,N_STATES,WEIGHT_Y \
+    --ncols 4 \
+    --n_chars_per_word 20 \
+    --n_words_per_topic 15 \
+    --rank_words_by $rank_words_by \
+    --show_longer_words_via_tooltip 1 \
+    --metrics_template_html "$template_html" \
+
+done
+
+
+popd

scripts/product_reviews/train_topic_models/rsync_snapshot_perf_csv.sh

@@ -0,0 +1,11 @@
+dataset_subpath="multi_domain_product_reviews_dataset/clean_data_v20180403/"
+dataset_path="$HOME/git/$dataset_subpath/"
+echo "dataset_path:"
+echo "$dataset_path"
+
+export XHOST_SSH_ADDR=mchughes@browncs
+export XHOST_REMOTE_PATH="/nbu/liv/mhughes/slda_results/$dataset_subpath/"
+export XHOST_LOCAL_PATH="/results/$dataset_subpath/"
+
+bash $PC_REPO_DIR/scripts/rsync_tools/rsync_snapshot_perf_metrics.sh
+

scripts/product_reviews/train_topic_models/select_best_snapshots.sh

@@ -0,0 +1,29 @@
+dataset_subpath="multi_domain_product_reviews_dataset/clean_data_v20180403/"
+dataset_path="$HOME/git/$dataset_subpath/"
+echo "dataset_path:"
+echo "$dataset_path"
+
+export XHOST_SSH_ADDR=mchughes@browncs
+export XHOST_REMOTE_PATH="/nbu/liv/mhughes/slda_results/$dataset_subpath/"
+export XHOST_LOCAL_PATH="/results/$dataset_subpath/"
+
+y_colnames=`cat $dataset_path/Y_colnames.txt`
+echo "target_y_name: $y_colnames"
+
+results_path_pattern_01="$XHOST_LOCAL_PATH/20180301*tensorflow*"
+output_path="$XHOST_LOCAL_PATH/best_runs_20180301_pcslda_tensorflow/"
+
+python $PC_REPO_DIR/pc_toolbox/utils_snapshots/select_best_runs_and_snapshots.py \
+    --output_path $output_path \
+    --legend_name PC_sLDA \
+    --results_path_patterns "$results_path_pattern_01" \
+    --txt_src_path $dataset_path \
+    --target_y_name $y_colnames \
+    --all_y_names $y_colnames \
+    --selection_score_colname Y_ROC_AUC \
+    --selection_score_ranking_func argmax \
+    --col_names_to_use_at_selection N_STATES,WEIGHT_Y \
+    --col_names_to_keep_per_split \
+        Y_ROC_AUC,Y_ERROR_RATE,LOGPDF_X_PERTOK,LOGPDF_Y_PERDOC \
+    --col_names_to_keep \
+        ALPHA,TAU,LAMBDA_W,N_BATCHES \