Adding SE score

global_func/predict_dupes.py

@@ -442,6 +442,8 @@ def predict_dupes(portfolio, maps_dict, site_var, type_var, Contest_Size, streng
     portfolio['Lineup Edge'] = 2 * max_edge * (portfolio['Lineup Edge'] - portfolio['Lineup Edge'].min()) / (portfolio['Lineup Edge'].max() - portfolio['Lineup Edge'].min()) - max_edge
     portfolio['Weighted Own'] = portfolio[own_columns].apply(calculate_weighted_ownership_wrapper, axis=1)
     portfolio['Geomean'] = np.power((portfolio[own_columns] * 100).product(axis=1), 1 / len(own_columns))
+    portfolio['SE Score'] = ((portfolio['median'] - portfolio['median'].mean()) * (portfolio['Weighted Own'] - portfolio['Weighted Own'].mean()))
+    portfolio['SE Score'] = (np.tanh(portfolio['SE Score'] / portfolio['SE Score'].std()) + 1) / 2
     
     # Calculate similarity score based on actual player selection
     portfolio['Diversity'] = calculate_player_similarity_score_chunked(portfolio, player_columns)
@@ -452,7 +454,7 @@ def predict_dupes(portfolio, maps_dict, site_var, type_var, Contest_Size, streng
     
     int16_columns_stacks = ['Dupes', 'Size', 'salary']
     int16_columns_nstacks = ['Dupes', 'salary']
-    float32_columns = ['median', 'Own', 'Finish_percentile', 'Win%', 'Lineup Edge', 'Weighted Own', 'Geomean', 'Diversity']
+    float32_columns = ['median', 'Own', 'Finish_percentile', 'Win%', 'SE Score', 'Lineup Edge', 'Weighted Own', 'Geomean', 'Diversity']
 
     try:
         portfolio[int16_columns_stacks] = portfolio[int16_columns_stacks].astype('uint16')