wsba-hockey 1.0.2__py3-none-any.whl → 1.0.3__py3-none-any.whl

wsba_hockey/tools/plotting.py

@@ -14,9 +14,9 @@ event_markers = {
     'faceoff':'X',
     'hit':'P',
     'blocked-shot':'v',
-    'missed-shot':'H',
+    'missed-shot':'o',
     'shot-on-goal':'D',
-    'goal':'o',
+    'goal':'*',
     'giveaway':'1',
     'takeaway':'2',
 }
@@ -135,7 +135,7 @@ def plot_game_events(pbp,game_id,events,strengths,marker_dict=event_markers,team
 
     for event in events:
         plays = pbp.loc[pbp['event_type']==event]
-        ax.scatter(plays['x_adj'],plays['y_adj'],plays['size'],plays['color'],marker=event_markers[event],label=event,zorder=5)
+        ax.scatter(plays['x_adj'],plays['y_adj'],plays['size'],plays['color'],marker=event_markers[event],edgecolors='white',label=event,zorder=5)
 
     ax.set_title(f'{away_abbr} @ {home_abbr} - {date}')
     ax.legend(bbox_to_anchor =(0.5,-0.35), loc='lower center',ncol=1).set_visible(legend)

wsba_hockey/tools/xg_model.py

@@ -108,7 +108,7 @@ def wsba_xG(pbp, train = False, overwrite = False, model_path = "tools/xg_model/
             'prior_faceoff']
     
     #Prep Data
-    pbp = prep_xG_data(pbp)
+    pbp_prep = prep_xG_data(pbp)
     #Filter unwanted date:
     #Shots must occur in specified events and strength states, occur before the shootout, and have valid coordinates
     events = ['faceoff','hit','giveaway','takeaway','blocked-shot','missed-shot','shot-on-goal','goal']
@@ -127,12 +127,12 @@ def wsba_xG(pbp, train = False, overwrite = False, model_path = "tools/xg_model/
                 '6v4',
                 '6v5']
     
-    data = pbp.loc[(pbp['event_type'].isin(events))&
-                   (pbp['strength_state'].isin(strengths))&
-                   (pbp['period'] < 5)&
-                   (pbp['x_fixed'].notna())&
-                   (pbp['y_fixed'].notna())&
-                   ~((pbp['x_fixed']==0)&(pbp['y_fixed']==0)&(pbp['x_fixed'].isin(fenwick_events))&(pbp['event_distance']!=90))]
+    data = pbp_prep.loc[(pbp_prep['event_type'].isin(events))&
+                   (pbp_prep['strength_state'].isin(strengths))&
+                   (pbp_prep['period'] < 5)&
+                   (pbp_prep['x_fixed'].notna())&
+                   (pbp_prep['y_fixed'].notna())&
+                   ~((pbp_prep['x_fixed']==0)&(pbp_prep['y_fixed']==0)&(pbp_prep['x_fixed'].isin(fenwick_events))&(pbp_prep['event_distance']!=90))]
 
     #Convert to sparse
     data_sparse = sp.csr_matrix(data[[target]+continous+boolean])
@@ -199,11 +199,11 @@ def wsba_xG(pbp, train = False, overwrite = False, model_path = "tools/xg_model/
         best_all = best_all.sort_values(by="auc", ascending=False)
 
         if overwrite == True:
-            best_all.to_csv("xg_model/testing/xg_model_training_runs.csv",index=False)
+            best_all.to_csv("tools/xg_model/testing/xg_model_training_runs.csv",index=False)
         else: 
-            best_old = pd.read_csv("xg_model/testing/xg_model_training_runs.csv")
+            best_old = pd.read_csv("tools/xg_model/testing/xg_model_training_runs.csv")
             best_comb = pd.concat([best_old,best_all])
-            best_comb.to_csv("xg_model/testing/xg_model_training_runs.csv",index=False)
+            best_comb.to_csv("tools/xg_model/testing/xg_model_training_runs.csv",index=False)
 
         # Final parameters
         param_7_EV = {
@@ -250,11 +250,11 @@ def wsba_xG(pbp, train = False, overwrite = False, model_path = "tools/xg_model/
         # Clean results and sort to find the number of rounds to use and seed
         cv_final = cv_test.sort_values(by="AUC", ascending=False)
         if overwrite == True:
-            cv_final.to_csv("xg_model/testing/xg_model_cv_runs.csv",index=False)
+            cv_final.to_csv("tools/xg_model/testing/xg_model_cv_runs.csv",index=False)
         else:
-            cv_old = pd.read_csv("xg_model/testing/xg_model_cv_runs.csv")
+            cv_old = pd.read_csv("tools/xg_model/testing/xg_model_cv_runs.csv")
             cv_comb = pd.concat([cv_old,cv_final])
-            cv_comb.to_csv("xg_model/testing/xg_model_cv_runs.csv")
+            cv_comb.to_csv("tools/xg_model/testing/xg_model_cv_runs.csv")
         cv_final.loc[len(cv_final)] = cv_test.mean()
 
         # Train the final model
@@ -276,8 +276,34 @@ def wsba_xG(pbp, train = False, overwrite = False, model_path = "tools/xg_model/
         
     else:
         model = joblib.load(model_path)
-        pbp['xG'] = np.where(pbp['event_type'].isin(fenwick_events),model.predict(xgb_matrix),"")
-        return pbp
+
+        #Predict goal
+        data['xG'] = model.predict(xgb_matrix)
+        data['xG'] = np.where(data['event_type'].isin(fenwick_events),data['xG'],np.nan)
+
+        #Avoid merging errors
+        merge_col = ['game_id','period','seconds_elapsed','event_type','event_team_abbr','event_player_1_id']
+        
+        for df in [pbp,data]:
+            df = df.astype({
+              'game_id':'int',
+              'period':'int',
+              'seconds_elapsed':'int',
+              'event_type':'str',
+              'event_team_abbr':'str',
+              'event_player_1_id':'float'
+            })
+
+        #Drop previous xG if it exists
+        try: pbp = pbp.drop(columns=['xG'])
+        except KeyError:
+            ''
+
+        #Merge
+        data = data[merge_col+['xG']]
+        pbp_xg = pd.merge(pbp,data,how='left')
+
+        return pbp_xg
 
 def moneypuck_xG(pbp,repo_path = "tools/xg_model/moneypuck/shots_2007-2023.zip"):
     #Given play-by-play, return itself with xG column sourced from MoneyPuck.com

wsba_hockey/workspace.py

@@ -8,6 +8,7 @@ from gspread_pandas import Spread, Client
 import urllib.request
 from wand.color import Color
 from wand.image import Image
+from tools.xg_model import wsba_xG
 
 season_load = wsba.repo_load_seasons()
 
@@ -21,6 +22,13 @@ def workspace(seasons,type,arg = '',start='',end=''):
             data['pbp'].to_csv(f'pbp/nhl_pbp_{season}.csv',index=False)
         print(f'Errors: {errors}')
 
+    elif type == 'pbp_xg':
+        #Add xG to pbp
+        for season in seasons:
+            print(f'WSBA xG for {season}')
+            data = pd.read_parquet(f'pbp/parquet/nhl_pbp_{season}.parquet')
+            wsba_xG(data).to_parquet(f'pbp/parquet/nhl_pbp_{season}.parquet',index=False)
+
     elif type == 'convert':
         for season in seasons:
             data = pd.read_csv(f"pbp/nhl_pbp_{season}.csv")
@@ -40,20 +48,26 @@ def workspace(seasons,type,arg = '',start='',end=''):
     
     elif type == 'stats':
         #Stats building
-        stats = []
-        for season in seasons:
-            for group in ['skater','team']:
+        for group in ['skater','team']:
+            stats = []
+            for season in seasons:
                 pbp = pd.read_parquet(f'pbp/parquet/nhl_pbp_{season}.parquet')
                 stat = wsba.nhl_calculate_stats(pbp,group,[2],['5v5'],shot_impact=True)
                 stat.to_csv(f'stats/{group}/wsba_nhl_{season}_{group}.csv',index=False)
-                stats.append(stat)
-            pd.concat(stats).to_csv(f'stats/db/wsba_nhl_{group}_db.csv',index=False)
-    
+                stats.append(stat) 
+
+            if arg:
+                pd.concat(stats).to_csv(f'stats/db/wsba_nhl_{group}_db.csv',index=False)
+
+    elif type == 'xg_model':
+        data = pd.concat([pd.read_parquet(f'pbp/parquet/nhl_pbp_{season}.parquet') for season in seasons])
+        wsba.wsba_main.wsba_xG(data,True,True)
+        
     elif type == 'plot_game':
         for season in seasons:
             pbp = wsba.nhl_scrape_season(season,remove=[],start=start,end=end)
 
-            plots = wsba.nhl_plot_games(pbp,wsba.wsba_main.fenwick_events,['5v5'],'all',team_colors=arg,legend=True)
+            plots = wsba.nhl_plot_games(pbp,wsba.wsba_main.fenwick_events,['5v5'],'all',team_colors=arg,legend=False,xg='wsba')
 
             games = list(pbp['game_id'].astype(str).drop_duplicates())
             i = 1
@@ -111,4 +125,6 @@ def push_to_sheet():
     spread.df_to_sheet(country,index=False,sheet='Country Info')
     spread.df_to_sheet(schedule,index=False,sheet='Schedule')
 
-workspace(['20242025'],'plot_game',arg={'away':'secondary','home':'primary'}, start='04-20',end='04-20')
+    print('Done.')
+
+workspace(season_load[6:18],,Tru)

wsba_hockey/wsba_main.py

@@ -541,15 +541,15 @@ def nhl_shooting_impacts(agg,team=False):
         pos['RushF/60'] = (pos['RushF']/pos['TOI'])*60
         pos['RushA/60'] = (pos['RushA']/pos['TOI'])*60
         pos['Rushes FF'] = pos['RushF/60'].rank(pct=True)
-        pos['Rushes FA'] = pos['RushA/60'].rank(pct=True)
+        pos['Rushes FA'] = 1 - pos['RushA/60'].rank(pct=True)
         pos['RushFxG/60'] = (pos['RushFxG']/pos['TOI'])*60
         pos['RushAxG/60'] = (pos['RushAxG']/pos['TOI'])*60
         pos['Rushes xGF'] = pos['RushFxG/60'].rank(pct=True)
-        pos['Rushes xGA'] = pos['RushAxG/60'].rank(pct=True)
+        pos['Rushes xGA'] = 1 - pos['RushAxG/60'].rank(pct=True)
         pos['RushFG/60'] = (pos['RushFG']/pos['TOI'])*60
         pos['RushAG/60'] = (pos['RushAG']/pos['TOI'])*60
         pos['Rushes GF'] = pos['RushFG/60'].rank(pct=True)
-        pos['Rushes GA'] = pos['RushAG/60'].rank(pct=True)
+        pos['Rushes GA'] = 1 - pos['RushAG/60'].rank(pct=True)
 
         #Flip against metric percentiles
         pos['ODEF-SR'] = 1-pos['ODEF-SR']
@@ -642,7 +642,7 @@ def nhl_calculate_stats(pbp,type,season_types,game_strength,roster_path="rosters
     # param 'xg' - xG model to apply to pbp for aggregation
     # param 'shot_impact' - boolean determining if the shot impact model will be applied to the dataset
 
-    print(f"Calculating statistics for all games in the provided play-by-play data...\nSeasons included: {pbp['season'].drop_duplicates().to_list()}...")
+    print(f"Calculating statistics for all games in the provided play-by-play data for {type}s...\nSeasons included: {pbp['season'].drop_duplicates().to_list()}...")
     start = time.perf_counter()
 
     #Add extra data and apply team changes
@@ -674,6 +674,9 @@ def nhl_calculate_stats(pbp,type,season_types,game_strength,roster_path="rosters
     if type == 'team':
         complete = calc_team(pbp)
 
+        #WSBA
+        complete['WSBA'] = complete['Team']+complete['Season'].astype(str)
+
         #Set TOI to minute
         complete['TOI'] = complete['TOI']/60
 
@@ -681,6 +684,14 @@ def nhl_calculate_stats(pbp,type,season_types,game_strength,roster_path="rosters
         for stat in per_sixty[7:13]:
             complete[f'{stat}/60'] = (complete[stat]/complete['TOI'])*60
 
+        #Rank per 60 stats
+        for stat in per_sixty[7:13]:
+            complete[f'{stat}/60 Percentile'] = complete[f'{stat}/60'].rank(pct=True)
+
+        #Flip percentiles for against stats
+        for stat in ['FA','xGA','GA']:
+            complete[f'{stat}/60 Percentile'] = 1-complete[f'{stat}/60 Percentile']
+
         end = time.perf_counter()
         length = end-start
         print(f'...finished in {(length if length <60 else length/60):.2f} {'seconds' if length <60 else 'minutes'}.')
@@ -735,6 +746,14 @@ def nhl_calculate_stats(pbp,type,season_types,game_strength,roster_path="rosters
         for stat in per_sixty:
             complete[f'{stat}/60'] = (complete[stat]/complete['TOI'])*60
 
+        #Rank per 60 stats
+        for stat in per_sixty:
+            complete[f'{stat}/60 Percentile'] = complete[f'{stat}/60'].rank(pct=True)
+
+        #Flip percentiles for against stats
+        for stat in ['FA','xGA','GA']:
+            complete[f'{stat}/60 Percentile'] = 1-complete[f'{stat}/60 Percentile']
+
         #Add player age
         complete['Birthday'] = pd.to_datetime(complete['Birthday'])
         complete['season_year'] = complete['Season'].astype(str).str[4:8].astype(int)
@@ -768,7 +787,7 @@ def nhl_calculate_stats(pbp,type,season_types,game_strength,roster_path="rosters
             "GF","FF","xGF","xGF/FF","GF/xGF","FshF%",
             "GA","FA","xGA","xGA/FA","GA/xGA","FshA%",
             'Rush',"Rush xG",'Rush G',"GC%","AC%","GI%","FC%","xGC%",
-        ]+[f'{stat}/60' for stat in per_sixty]+type_metrics].fillna(0).sort_values(['Player','Season','Team','ID'])
+        ]+[f'{stat}/60' for stat in per_sixty]+[f'{stat}/60 Percentile' for stat in per_sixty]+type_metrics].fillna(0).sort_values(['Player','Season','Team','ID'])
         
         print(f'...finished in {(length if length <60 else length/60):.2f} {'seconds' if length <60 else 'minutes'}.')
         #Apply shot impacts if necessary (Note: this will remove skaters with fewer than 150 minutes of TOI due to the shot impact TOI rule)

{wsba_hockey-1.0.2.dist-info → wsba_hockey-1.0.3.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: wsba_hockey
-Version: 1.0.2
+Version: 1.0.3
 Summary: WeakSide Breakout's complete Python package of access to hockey data, primairly including the scraping of National Hockey League schedule, play-by-play, and shifts information.
 Author-email: Owen Singh <owenbksingh@gmail.com>
 Project-URL: Homepage, https://github.com/owensingh38/wsba_hockey/
@@ -69,7 +69,7 @@ wsba.nhl_scrape_prospects('BOS')
 ### Stat Aggregation
 ```python
 pbp = wsba.nhl_scrape_season('20232024',remove=[], local = True)
-wsba.nhl_calculate_stats(pbp,'skater',[2],['5v5','4v4','3v3'],xg='moneypuck',shot_impact = True)
+wsba.nhl_calculate_stats(pbp,'skater',[2],['5v5','4v4','3v3'],xg='wsba',shot_impact = True)
 ```
 ### Shot Plotting (Plots, Heatmaps, etc.)
 ```python

{wsba_hockey-1.0.2.dist-info → wsba_hockey-1.0.3.dist-info}/RECORD

@@ -1,19 +1,19 @@
 wsba_hockey/__init__.py,sha256=QXyc8FFlCDWQuECyyEbj80ASHEbTFj4R13DOFOY9nJg,353
-wsba_hockey/workspace.py,sha256=HJ4ZJyL8OwrtknXKAqiptW8WxbJZp3kc0bobDwpKcLY,4875
-wsba_hockey/wsba_main.py,sha256=ZYfYpqJeUBnVHAjGrDWkEjoYW_8qWCilYukGuKUkJA4,37073
+wsba_hockey/workspace.py,sha256=YQMHCAiCCsJiaSs_MZI_2tKuQp6dmQImIZw-RvjBEhA,5395
+wsba_hockey/wsba_main.py,sha256=xyZLnOZxzPAi1-n9mBsQ8ThyGJkVotRg5I39SiM_CYs,37886
 wsba_hockey/stats/calculate_viz/shot_impact.py,sha256=7zxf64yt87YDucUBG31W75u951AUbMC7a3x5ClNIxYI,39
 wsba_hockey/tools/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 wsba_hockey/tools/agg.py,sha256=WRfuxJt0OgDSEkqHtuGql9nZrQnLGzkm6DUFVbDXayE,9560
-wsba_hockey/tools/plotting.py,sha256=mOkXO-ZWFTblsGpYSOZFxFJFad4sZBr62prVaatoLEw,6003
+wsba_hockey/tools/plotting.py,sha256=olPGYJtVFIjBwrOQyz1CNNqaCM6iOJTKjWXBA3egJHM,6022
 wsba_hockey/tools/scraping.py,sha256=hZv1XMtQjsVSSVwN6Fzw7FuT94zSQGyX1WupTjUjUuU,48548
-wsba_hockey/tools/xg_model.py,sha256=vdmDq9CWFr-2AKUrnorjUyweUf9_NMdq5Xf3mbPserw,15158
+wsba_hockey/tools/xg_model.py,sha256=ItVWGvYzi3zHA8mPCkUbpTU7NUcQFSw3Xm_nV0E6ypQ,16004
 wsba_hockey/tools/archive/old_scraping.py,sha256=hEjMI1RtfeZnf0RBiJFI38oXkLZ3WofeH5xqcF4pzgM,49585
 wsba_hockey/tools/utils/__init__.py,sha256=vccXhOtzARoR99fmEWU1OEI3qCIdQ9Z42AlRA_BUhrs,114
 wsba_hockey/tools/utils/config.py,sha256=D3Uk05-YTyrhfReMTTLfNI3HN_rON2uo_CDE9oER3Lg,351
 wsba_hockey/tools/utils/save_pages.py,sha256=CsyL_0n-b-4pJoUauwU3HpnCO6n69-RlBMJQBd_qGDc,4979
 wsba_hockey/tools/utils/shared.py,sha256=dH_JwZfia5fib8rksy5sW-mBp0pluBPvw37Vdr8Kap0,14211
-wsba_hockey-1.0.2.dist-info/licenses/LICENSE,sha256=Nr_Um1Pd5FQJTWWgm7maZArdtYMbDhzXYSwyJIZDGik,1114
-wsba_hockey-1.0.2.dist-info/METADATA,sha256=fBFZblpMy7vFKcM9MN4JBjJYSIIljdfVNd4GusrpbKU,3547
-wsba_hockey-1.0.2.dist-info/WHEEL,sha256=pxyMxgL8-pra_rKaQ4drOZAegBVuX-G_4nRHjjgWbmo,91
-wsba_hockey-1.0.2.dist-info/top_level.txt,sha256=acU7s3x-RZC1zGiqCOmO0g267iqCg34lzIfdmYxxGmQ,12
-wsba_hockey-1.0.2.dist-info/RECORD,,
+wsba_hockey-1.0.3.dist-info/licenses/LICENSE,sha256=Nr_Um1Pd5FQJTWWgm7maZArdtYMbDhzXYSwyJIZDGik,1114
+wsba_hockey-1.0.3.dist-info/METADATA,sha256=Cds728R4Mz7RqNuOA0mnMciv4wPUjkNShlPScxaKclw,3542
+wsba_hockey-1.0.3.dist-info/WHEEL,sha256=pxyMxgL8-pra_rKaQ4drOZAegBVuX-G_4nRHjjgWbmo,91
+wsba_hockey-1.0.3.dist-info/top_level.txt,sha256=acU7s3x-RZC1zGiqCOmO0g267iqCg34lzIfdmYxxGmQ,12
+wsba_hockey-1.0.3.dist-info/RECORD,,