bszek213 · bszek213 · Nov 14, 2022 · Nov 14, 2022 · Nov 15, 2022 · Nov 15, 2022
diff --git a/.gitignore b/.gitignore
@@ -2,7 +2,8 @@
 __pycache__/
 *.py[cod]
 *$py.class
-
+#Saved models
+randomForestModelTuned.joblib
 # C extensions
 *.so
 

diff --git a/FeatureImportance.png b/FeatureImportance.png
diff --git a/__pycache__/nba_web_scraper.cpython-310.pyc b/__pycache__/nba_web_scraper.cpython-310.pyc
diff --git a/all_data_regressor.csv b/all_data_regressor.csv
diff --git a/correlations.png b/correlations.png
diff --git a/nba_ml.py b/nba_ml.py
diff --git a/nba_web_scraper.py b/nba_web_scraper.py
@@ -13,22 +13,28 @@
 from os.path import join, exists
 from os import getcwd
 
-def html_to_df_web_scrape_NBA(URL,team,year):
+def html_to_df_web_scrape_NBA(URL,URL1,team,year):
     while True:
         try:
             page = requests.get(URL)
             soup = BeautifulSoup(page.content, "html.parser")
+            page1 = requests.get(URL1)
+            soup1 = BeautifulSoup(page1.content, "html.parser")
             break
         except:
             print('HTTPSConnectionPool(host="www.sports-reference.com", port=443): Max retries exceeded. Retry in 10 seconds')
             sleep(10)
     table = soup.find(id="all_tgl_basic")
+    table1 = soup1.find(id="all_tgl_advanced")
     tbody = table.find('tbody')
+    tbody1 = table1.find('tbody')
     tr_body = tbody.find_all('tr')
+    tr_body1 = tbody1.find_all('tr')
     # game_season = []
     # date_game = []
     # game_location = []
     # opp_id= []
+    # BASIC STATS
     game_result= []
     pts= []
     opp_pts= []
@@ -138,14 +144,77 @@ def html_to_df_web_scrape_NBA(URL,team,year):
             if td.get('data-stat') == "opp_tov":
                 opp_tov.append(td.get_text())
             if td.get('data-stat') == "opp_pf":
-                opp_pf.append(td.get_text())          
+                opp_pf.append(td.get_text())      
+    #ADVANCED STATS
+    off_rtg = []
+    def_rtg = []
+    pace = []
+    fta_per_fga_pct = []
+    fg3a_per_fga_pct = []
+    ts_pct = []
+    trb_pct = []
+    ast_pct = []
+    stl_pct = [] 
+    blk_pct = []
+    efg_pct = []
+    tov_pct = []
+    orb_pct = []
+    ft_rate = []
+    opp_efg_pct= []
+    opp_tov_pct = []
+    drb_pct = []
+    opp_ft_rate = []
+    for trb in tr_body1:
+        for td in trb.find_all('td'):
+            if td.get('data-stat') == "off_rtg":
+                off_rtg.append(td.get_text())
+            if td.get('data-stat') == "def_rtg":
+                def_rtg.append(td.get_text())
+            if td.get('data-stat') == "pace":
+                pace.append(td.get_text())
+            if td.get('data-stat') == "fta_per_fga_pct":
+                fta_per_fga_pct.append(td.get_text())
+            if td.get('data-stat') == "fg3a_per_fga_pct":
+                fg3a_per_fga_pct.append(td.get_text())
+            if td.get('data-stat') == "ts_pct":
+                ts_pct.append(td.get_text())
+            if td.get('data-stat') == "trb_pct":
+                trb_pct.append(td.get_text())
+            if td.get('data-stat') == "ast_pct":
+                ast_pct.append(td.get_text())
+            if td.get('data-stat') == "stl_pct":
+                stl_pct.append(td.get_text())
+            if td.get('data-stat') == "blk_pct":
+                blk_pct.append(td.get_text())
+            if td.get('data-stat') == "efg_pct":
+                efg_pct.append(td.get_text())
+            if td.get('data-stat') == "tov_pct":
+                tov_pct.append(td.get_text())
+            if td.get('data-stat') == "orb_pct":
+                orb_pct.append(td.get_text())
+            if td.get('data-stat') == "ft_rate":
+                ft_rate.append(td.get_text())
+            if td.get('data-stat') == "opp_efg_pct":
+                opp_efg_pct.append(td.get_text())
+            if td.get('data-stat') == "opp_tov_pct":
+                opp_tov_pct.append(td.get_text())
+            if td.get('data-stat') == "drb_pct":
+                drb_pct.append(td.get_text())
+            if td.get('data-stat') == "opp_ft_rate":
+                opp_ft_rate.append(td.get_text())
+
+
     return DataFrame(list(zip(game_result,pts,opp_pts,fg,fga,
     fg_pct,fg3,fg3a,fg3_pct,ft,fta,ft_pct,orb,total_board,ast,
     stl,blk,tov,pf,opp_fg,opp_fga,opp_fg_pct,opp_fg3,opp_fg3a,opp_fg3_pct,
     opp_ft,opp_fta,opp_ft_pct,opp_orb,opp_trb,opp_ast,opp_stl,opp_blk,opp_tov,
-    opp_pf)),
+    opp_pf, off_rtg,def_rtg,pace,fta_per_fga_pct,fg3a_per_fga_pct,ts_pct,
+    trb_pct,ast_pct,stl_pct,blk_pct,efg_pct,tov_pct,orb_pct,ft_rate,opp_efg_pct,
+    opp_tov_pct,drb_pct,opp_ft_rate)),
             columns =['game_result','pts','opp_pts','fg','fga',
             'fg_pct','fg3','fg3a','fg3_pct','ft','fta','ft_pct','orb','total_board','ast',
             'stl','blk','tov','pf','opp_fg','opp_fga','opp_fg_pct','opp_fg3','opp_fg3a','opp_fg3_pct',
             'opp_ft','opp_fta','opp_ft_pct','opp_orb','opp_trb','opp_ast','opp_stl','opp_blk','opp_tov',
-            'opp_pf'])
+            'opp_pf','off_rtg','def_rtg','pace','fta_per_fga_pct','fg3a_per_fga_pct','ts_pct',
+            'trb_pct','ast_pct','stl_pct','blk_pct','efg_pct','tov_pct','orb_pct','ft_rate','opp_efg_pct',
+            'opp_tov_pct','drb_pct','opp_ft_rate'])
diff --git a/prob_plots_regress/probplot_ast.png b/prob_plots_regress/probplot_ast.png
diff --git a/prob_plots_regress/probplot_ast_pct.png b/prob_plots_regress/probplot_ast_pct.png
diff --git a/prob_plots_regress/probplot_blk.png b/prob_plots_regress/probplot_blk.png
diff --git a/prob_plots_regress/probplot_fg.png b/prob_plots_regress/probplot_fg.png
diff --git a/prob_plots_regress/probplot_fg3.png b/prob_plots_regress/probplot_fg3.png
diff --git a/prob_plots_regress/probplot_fg3_pct.png b/prob_plots_regress/probplot_fg3_pct.png
diff --git a/prob_plots_regress/probplot_fg_pct.png b/prob_plots_regress/probplot_fg_pct.png
diff --git a/prob_plots_regress/probplot_fga.png b/prob_plots_regress/probplot_fga.png
diff --git a/prob_plots_regress/probplot_ft.png b/prob_plots_regress/probplot_ft.png
diff --git a/prob_plots_regress/probplot_ft_pct.png b/prob_plots_regress/probplot_ft_pct.png
diff --git a/prob_plots_regress/probplot_game_result.png b/prob_plots_regress/probplot_game_result.png
diff --git a/prob_plots_regress/probplot_off_rtg.png b/prob_plots_regress/probplot_off_rtg.png
diff --git a/prob_plots_regress/probplot_opp_ast.png b/prob_plots_regress/probplot_opp_ast.png
diff --git a/prob_plots_regress/probplot_opp_blk.png b/prob_plots_regress/probplot_opp_blk.png
diff --git a/prob_plots_regress/probplot_opp_fg3.png b/prob_plots_regress/probplot_opp_fg3.png
diff --git a/prob_plots_regress/probplot_opp_fg3_pct.png b/prob_plots_regress/probplot_opp_fg3_pct.png
diff --git a/prob_plots_regress/probplot_opp_fg_pct.png b/prob_plots_regress/probplot_opp_fg_pct.png
diff --git a/prob_plots_regress/probplot_opp_fga.png b/prob_plots_regress/probplot_opp_fga.png
diff --git a/prob_plots_regress/probplot_opp_ft.png b/prob_plots_regress/probplot_opp_ft.png
diff --git a/prob_plots_regress/probplot_opp_ft_pct.png b/prob_plots_regress/probplot_opp_ft_pct.png
diff --git a/prob_plots_regress/probplot_opp_orb.png b/prob_plots_regress/probplot_opp_orb.png
diff --git a/prob_plots_regress/probplot_opp_pf.png b/prob_plots_regress/probplot_opp_pf.png
diff --git a/prob_plots_regress/probplot_opp_pts.png b/prob_plots_regress/probplot_opp_pts.png
diff --git a/prob_plots_regress/probplot_opp_stl.png b/prob_plots_regress/probplot_opp_stl.png
diff --git a/prob_plots_regress/probplot_opp_tov.png b/prob_plots_regress/probplot_opp_tov.png
diff --git a/prob_plots_regress/probplot_opp_trb.png b/prob_plots_regress/probplot_opp_trb.png
diff --git a/prob_plots_regress/probplot_orb.png b/prob_plots_regress/probplot_orb.png
diff --git a/prob_plots_regress/probplot_pace.png b/prob_plots_regress/probplot_pace.png
diff --git a/prob_plots_regress/probplot_pf.png b/prob_plots_regress/probplot_pf.png
diff --git a/prob_plots_regress/probplot_stl.png b/prob_plots_regress/probplot_stl.png
diff --git a/prob_plots_regress/probplot_total_board.png b/prob_plots_regress/probplot_total_board.png
diff --git a/prob_plots_regress/probplot_tov.png b/prob_plots_regress/probplot_tov.png
diff --git a/prob_plots_regress/probplot_trb_pct.png b/prob_plots_regress/probplot_trb_pct.png
diff --git a/year_count.yaml b/year_count.yaml