Update stuff_model/calculate_arm_angles.py

stuff_model/calculate_arm_angles.py

@@ -1,72 +1,121 @@
-import polars as pl
-import numpy as np
-import requests
-
-def calculate_arm_angles(df: pl.DataFrame,pitcher_id:int) -> pl.DataFrame:
-    df_arm_angle = pl.read_csv('stuff_model/pitcher_arm_angles_2024.csv')
-    #pitcher_id = 489446
-    df_filter = df.filter(pl.col("pitcher_id") == pitcher_id).drop_nulls(subset=["release_pos_x", "release_pos_z"])
-    # data = requests.get(f'https://statsapi.mlb.com/api/v1/people?personIds={pitcher_id}').json()
-
-    if pitcher_id not in df_arm_angle["pitcher"]:
-
-        data = requests.get(f'https://statsapi.mlb.com/api/v1/people?personIds={pitcher_id}').json()
-        height_in = data['people'][0]['height']
-        height = int(height_in.split("'")[0]) * 12 + int(height_in.split("'")[1].split('"')[0])
-        df_filter = (df_filter.with_columns(
-                (pl.col("release_pos_x") * 12).alias("release_pos_x"),
-                (pl.col("release_pos_z") * 12).alias("release_pos_z"),
-                (pl.lit(height * 0.70)).alias("shoulder_pos"),
-            )
-            .with_columns(
-                (pl.col("release_pos_z") - pl.col("shoulder_pos")).alias("Opp"),
-                pl.col("release_pos_x").abs().alias("Adj"),
-            )
-            .with_columns(
-                pl.struct(["Opp", "Adj"]).map_elements(lambda x: np.arctan2(x["Opp"], x["Adj"])).alias("arm_angle_rad")
-            ))
-
-        df_filter = (df_filter.with_columns(    
-            
-                pl.col("arm_angle_rad").degrees().alias("arm_angle")
-            
-            #.drop(["Opp", "arm_angle_rad"])
-            ))
-            
-    else:
-        shoulder_x = df_arm_angle.filter(pl.col("pitcher") == pitcher_id)["relative_shoulder_x"][0]
-        shoulder_z = df_arm_angle.filter(pl.col("pitcher") == pitcher_id)["shoulder_z"][0]
-        rel_x = df_arm_angle.filter(pl.col("pitcher") == pitcher_id)["relative_release_ball_x"][0]
-        rel_z = df_arm_angle.filter(pl.col("pitcher") == pitcher_id)["release_ball_z"][0]
-
-
-        ball_angle = df_arm_angle.filter(pl.col("pitcher") == pitcher_id)["ball_angle"][0]
-
-        hyp = np.sqrt((rel_x - shoulder_x)**2 + (rel_z - shoulder_z)**2)
-
-        print(shoulder_x, shoulder_z)
-
-        df_filter = (df_filter.with_columns(
-
-            )
-            .with_columns(
-                (pl.col("release_pos_z") - shoulder_z).alias("Opp"),
-                (pl.lit(hyp)).alias("Hyp"),
-            )
-            .with_columns(
-                pl.struct(["Opp","Hyp"]).map_elements(lambda x: np.arcsin(x["Opp"] / x["Hyp"])).alias("arm_angle_rad")
-            )
-            .with_columns(
-                pl.col("arm_angle_rad").degrees().alias("arm_angle")
-            )
-            #.drop(["Opp", "arm_angle_rad"])
-            )
-    
-        df_filter = df_filter.with_columns(
-                ((pl.col("arm_angle") * 0.5) + (ball_angle * 0.5)).alias("arm_angle")
-            )
-    
-
-
-    
+import polars as pl
+import numpy as np
+import requests
+from io import StringIO
+
+def calculate_arm_angles(df: pl.DataFrame,pitcher_id:int) -> pl.DataFrame:
+
+    url = f"https://baseballsavant.mlb.com/leaderboard/pitcher-arm-angles?batSide=&dateStart={df['game_date'][0]}&dateEnd={df['game_date'][-1]}&gameType=R&groupBy=&min=1&minGroupPitches=1&perspective=back&pitchHand=&pitchType=&season={int(df['game_date'][0][0:4])}&size=small&sort=ascending&team=&csv=true"
+    
+    headers = {
+        'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/58.0.3029.110 Safari/537.3'
+    }
+    
+    response = requests.get(url, headers=headers)
+
+    old_data = False
+    daily_check = df['game_date'][0] == df['game_date'][-1]
+    
+    # Assuming response.text contains the CSV formatted string
+    csv_data = response.text
+    
+    # Use StringIO to convert the string into a file-like object
+    data = StringIO(csv_data)
+    
+    # Read the CSV data into a DataFrame
+    df_arm_angle = pl.read_csv(data)
+    
+    if pitcher_id not in df_arm_angle["pitcher"]:
+        old_data = True
+        df_arm_angle = pl.read_csv('stuff_model/pitcher_arm_angles_2024.csv')
+
+        
+    #pitcher_id = 489446
+    df_filter = df.filter(pl.col("pitcher_id") == pitcher_id).drop_nulls(subset=["release_pos_x", "release_pos_z"])
+    # data = requests.get(f'https://statsapi.mlb.com/api/v1/people?personIds={pitcher_id}').json()
+
+    if pitcher_id not in df_arm_angle["pitcher"]:
+
+        data = requests.get(f'https://statsapi.mlb.com/api/v1/people?personIds={pitcher_id}').json()
+        height_in = data['people'][0]['height']
+        height = int(height_in.split("'")[0]) * 12 + int(height_in.split("'")[1].split('"')[0])
+        df_filter = (df_filter.with_columns(
+                (pl.col("release_pos_x") * 12).alias("release_pos_x"),
+                (pl.col("release_pos_z") * 12).alias("release_pos_z"),
+                (pl.lit(height * 0.70)).alias("shoulder_pos"),
+            )
+            .with_columns(
+                (pl.col("release_pos_z") - pl.col("shoulder_pos")).alias("Opp"),
+                pl.col("release_pos_x").abs().alias("Adj"),
+            )
+            .with_columns(
+                pl.struct(["Opp", "Adj"]).map_elements(lambda x: np.arctan2(x["Opp"], x["Adj"])).alias("arm_angle_rad")
+            ))
+
+        df_filter = (df_filter.with_columns(    
+            
+                pl.col("arm_angle_rad").degrees().alias("arm_angle")
+            
+            #.drop(["Opp", "arm_angle_rad"])
+            ))
+            
+    else:
+        shoulder_x = df_arm_angle.filter(pl.col("pitcher") == pitcher_id)["relative_shoulder_x"][0]
+        shoulder_z = df_arm_angle.filter(pl.col("pitcher") == pitcher_id)["shoulder_z"][0]
+        rel_x = df_arm_angle.filter(pl.col("pitcher") == pitcher_id)["relative_release_ball_x"][0]
+        rel_z = df_arm_angle.filter(pl.col("pitcher") == pitcher_id)["release_ball_z"][0]
+
+
+        ball_angle = df_arm_angle.filter(pl.col("pitcher") == pitcher_id)["ball_angle"][0]
+
+        hyp = np.sqrt((rel_x - shoulder_x)**2 + (rel_z - shoulder_z)**2)
+
+        print(shoulder_x, shoulder_z)
+
+        df_filter = (df_filter.with_columns(
+
+            )
+            .with_columns(
+                (pl.col("release_pos_z") - shoulder_z).alias("Opp"),
+                (pl.lit(hyp)).alias("Hyp"),
+            )
+            .with_columns(
+                pl.struct(["Opp","Hyp"]).map_elements(lambda x: np.arcsin(x["Opp"] / x["Hyp"])).alias("arm_angle_rad")
+            )
+            .with_columns(
+                pl.col("arm_angle_rad").degrees().alias("arm_angle")
+            )
+            #.drop(["Opp", "arm_angle_rad"])
+            )
+        
+        if old_data:
+            df_filter = df_filter.with_columns(
+                    ((pl.col("arm_angle") * 0.5) + (ball_angle * 0.5)).alias("arm_angle")
+                )
+            
+        elif daily_check:
+            df_filter = df_filter.with_columns(
+                    ((pl.col("arm_angle") * 0.25) + (ball_angle * 0.75)).alias("arm_angle")
+                )
+            
+        else:
+            df_filter = df_filter.with_columns(
+                    ((pl.col("arm_angle") * 0.0) + (ball_angle * 1)).alias("arm_angle")
+                )        
+
+        valid_mean = df_filter["arm_angle"].fill_nan(None).drop_nulls().mean()
+
+        
+        df_filter = df_filter.with_columns(
+            df_filter["arm_angle"]
+            .fill_nan(None)  # Convert NaN to null
+            .fill_null(valid_mean)  # Fill nulls with mean
+        )
+   
+
+        #print([x for x in df_filter["arm_angle"]])
+                
+
+
+    
     return df_filter