Update app.py

app.py

@@ -3,15 +3,16 @@ import numpy as np
 import pandas as pd
 import pickle
 import xgboost as xgb
+from catboost import CatBoostRegressor
 
 def predict(team, inning, venue, hits, errors, lob, runs, opp_team, opp_runs, opp_hits):
     data = [team, opp_team, inning, venue, hits, opp_hits, errors, runs, opp_runs, lob]
     df = pd.DataFrame([data], columns=["Team_Name", "Opposition_Team", "Inning", "Home/Away", "Hits", "Opp_Hits", "Errors", "Runs", "Opp_Runs", "LOB"])
 
     xgb_model = xgb.XGBRegressor()
-    xgb_model.load_model('xgbr3_model3_exp8.json')
+    xgb_model.load_model('alpha1_xgbr1.json')
     
-    with open('label_encoder_teams_exp8.pkl', 'rb') as f:
+    with open('label_encoder_teams_alpha1_xgbr1.pkl', 'rb') as f:
         label_encoder = pickle.load(f)
 
     df['Team_Name'] = label_encoder.transform(df['Team_Name'])
@@ -20,6 +21,30 @@ def predict(team, inning, venue, hits, errors, lob, runs, opp_team, opp_runs, op
     home_away_status = {'Home': 0, 'Away': 1}
     df['Home/Away'] = df['Home/Away'].map(home_away_status)
     score = xgb_model.predict(df)
+    if score[0] < 0:
+        score = np.clip(score[0], a_min=0, a_max=None)
+        return np.round(score,1)
+    return np.round(score[0],1)
+
+def predict_2(team, inning, venue, hits, errors, lob, runs, opp_team, opp_runs, opp_hits):
+    data = [team, opp_team, inning, venue, hits, opp_hits, errors, runs, opp_runs, lob]
+    df = pd.DataFrame([data], columns=["Team_Name", "Opposition_Team", "Inning", "Home/Away", "Hits", "Opp_Hits", "Errors", "Runs", "Opp_Runs", "LOB"])
+
+    cat_model = CatBoostRegressor()
+    cat_model.load_model('CATBR1_model1_exp1.json')
+    
+    with open('label_encoder_teams_catbr1_exp1.pkl', 'rb') as f:
+        label_encoder = pickle.load(f)
+
+    df['Team_Name'] = label_encoder.transform(df['Team_Name'])
+    df['Opposition_Team'] = label_encoder.transform(df['Opposition_Team'])
+
+    home_away_status = {'Home': 0, 'Away': 1}
+    df['Home/Away'] = df['Home/Away'].map(home_away_status)
+    score = cat_model.predict(df)
+    if score[0] < 0:
+        score = np.clip(score[0], a_min=0, a_max=None)
+        return np.round(score,1)
     return np.round(score[0],1)
 
 team_names = ["Arizona Diamondbacks",
@@ -71,17 +96,30 @@ with gr.Blocks() as demo:
     #             </center>
     #             """)
     with gr.Row():
-        team = gr.Dropdown(choices = team_names, max_choices = 1, label="Team", scale=1)  
+        inning = gr.Number(None, label="Inning", minimum = 1, maximum = 8, scale=1)
+    
+    with gr.Row():
+        with gr.Column():
+            venue = gr.Dropdown(choices = ["Home", "Away"], value="Away", max_choices = 1, label="Home/Away Status", scale=1)  
+
         
         with gr.Column():
-            inning = gr.Number(None, label="Inning", minimum = 1, maximum = 8, scale=1)
+            opp_venue = gr.Dropdown(choices = ["Home", "Away"], value="Home", max_choices = 1, label="Opposition Home/Away Status", scale=1)  
+
     
     with gr.Row():
         with gr.Column():
-            venue = gr.Dropdown(choices = ["Home", "Away"], max_choices = 1, label="Home/Away Status", scale=1)  
-            
+            team = gr.Dropdown(choices = team_names, max_choices = 1, label="Team", scale=1)  
+        
+        with gr.Column():
+            opp_team = gr.Dropdown(choices = team_names, max_choices = 1, label="Opposition Team", scale=1) 
+    
+    with gr.Row():
         with gr.Column():
             hits = gr.Number(None, minimum=0, label="Hits - (H)", scale=1)
+            
+        with gr.Column():
+            opp_hits = gr.Number(None, minimum=0, label="Opposition Hits - (H)", scale=1)
     
     # summarize_btn = gr.Button(value="Summarize Text", size = 'sm')
 
@@ -90,30 +128,46 @@ with gr.Blocks() as demo:
             errors = gr.Number(None, minimum=0, label="Errors - (E)", scale=2)
         
         with gr.Column():
-            lob = gr.Number(None, minimum=0, label="Left on Base - (LOB)", scale=1)
+            opp_errors = gr.Number(None, minimum=0, label="Opposition Errors - (E)", scale=2)
 
         # runs = gr.Number(None, minimum=0, label="Runs - (R)", scale=1)
 
     with gr.Row():
         with gr.Column():
-            runs = gr.Number(None, minimum=0, label="Runs - (R)", scale=1)
-               
+            lob = gr.Number(None, minimum=0, label="Left on Base - (LOB)", scale=1)
+
         with gr.Column():
-            opp_team = gr.Dropdown(choices = team_names, max_choices = 1, label="Opposition Team", scale=1) 
+            opp_lob = gr.Number(None, minimum=0, label="Opposition Left on Base - (LOB)", scale=1)
 
     with gr.Row():
         with gr.Column():
-            opp_runs = gr.Number(None, minimum=0, label="Opposition Runs - (R)", scale=1)
-               
+            runs = gr.Number(None, minimum=0, label="Runs - (R)", scale=1)
+
         with gr.Column():
-            opp_hits = gr.Number(None, minimum=0, label="Opposition Hits - (H)", scale=1)
+            opp_runs = gr.Number(None, minimum=0, label="Opposition Runs - (R)", scale=1)
 
     with gr.Row():
         predict_btn = gr.Button(value="Predict", size = 'sm')
 
     with gr.Row():
-        final_score = gr.Textbox(label="Predicted Score", scale=1)
+        with gr.Column():
+            final_score_away1 = gr.Textbox(label="Predicted Score Model XGB", scale=1)
+
+        with gr.Column():
+            final_score_home1 = gr.Textbox(label="Opposition Predicted Score Model XGB", scale=1)
+
+    with gr.Row():
+        with gr.Column():
+            final_score_away2 = gr.Textbox(label="Predicted Score Model CATB", scale=1)
+
+        with gr.Column():
+            final_score_home2 = gr.Textbox(label="Opposition Predicted Score Model CATB", scale=1)
+
     # patent_doc.upload(document_to_text, inputs = [patent_doc, slider, select_model], outputs=summary_doc)
-    predict_btn.click(predict, inputs=[team, inning, venue, hits, errors, lob, runs, opp_team, opp_runs, opp_hits], outputs=final_score)
+    predict_btn.click(predict, inputs=[team, inning, venue, hits, errors, lob, runs, opp_team, opp_runs, opp_hits], outputs=final_score_away1)
+    predict_btn.click(predict, inputs=[opp_team, inning, opp_venue, opp_hits, opp_errors, opp_lob, opp_runs, team, runs, hits], outputs=final_score_home1)
+
+    predict_btn.click(predict_2, inputs=[team, inning, venue, hits, errors, lob, runs, opp_team, opp_runs, opp_hits], outputs=final_score_away2)
+    predict_btn.click(predict_2, inputs=[opp_team, inning, opp_venue, opp_hits, opp_errors, opp_lob, opp_runs, team, runs, hits], outputs=final_score_home2)
 
 demo.launch(inline=False)