nfelodcm 0.1.7__tar.gz → 0.1.9__tar.gz

{nfelodcm-0.1.7 → nfelodcm-0.1.9}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: nfelodcm
-Version: 0.1.7
+Version: 0.1.9
 Summary: Python package for loading and caching CSVs hosted on github into pandas dataframes
 Author-email: Robert Greer <nfl@robbygreer.com>
 Maintainer-email: Robert Greer <nfl@robbygreer.com>

nfelodcm-0.1.9/__init__.py

@@ -0,0 +1,5 @@
+from .nfelodcm.nfelodcm import get_df
+from .nfelodcm.nfelodcm import load
+from .nfelodcm.nfelodcm import get_map
+from .nfelodcm.nfelodcm import get_season_state
+from .nfelodcm.tests import test_all_maps

{nfelodcm-0.1.7 → nfelodcm-0.1.9}/nfelodcm/Engine/Assignments/assignment.py

@@ -1,11 +1,12 @@
 from .assignments import (
-    game_id_repl, penalty_formatting, desc_based_plays
+    fastr_team_id_repl, penalty_formatting, desc_based_plays,
+    score_clean
 )
 
 ## a dictionary that holds assignments and the columns they add ##
 assignments = {
-    'game_id_repl' : {
-        'func' : game_id_repl,
+    'fastr_team_id_repl' : {
+        'func' : fastr_team_id_repl,
         'new_columns' : []
     },
     'penalty_formatting' : {
@@ -19,6 +20,10 @@ assignments = {
             ('rush_attempt_all', 'float32'),
             ('play_call', 'str'),
         ]
+    },
+    'score_clean' : {
+        'func' : score_clean,
+        'new_columns' : []
     }
 }
 

nfelodcm-0.1.9/nfelodcm/Engine/Assignments/assignments/__init__.py

@@ -0,0 +1,3 @@
+from .fastr_team_repl import team_id_repl as fastr_team_id_repl
+from .penalty_formatting import penalty_formatting
+from .desc_based_plays import desc_based_plays

nfelodcm-0.1.9/nfelodcm/Engine/Assignments/assignments/fastr_team_repl.py

@@ -0,0 +1,34 @@
+import pandas as pd
+
+## assignment functions that replaces fastr team names with legacy team names ##
+repl = {
+    'LV' : 'OAK',
+    'SD' : 'LAC',
+    'STL' : 'LAR',
+    'LA' : 'LAR',
+}
+
+def team_id_repl(df):
+    """
+    Replaces fastr team ids with a legacy nfelo ids.
+    """
+    ## if a col with team names exists, replace it ##
+    for col in [
+        'home_team', 'away_team', 'team_abbr',
+        'posteam', 'defteam', 'penalty_team',
+        'side_of_field', 'timeout_team', 'td_team',
+        'return_team', 'possession_team',
+        'recent_team', 'opponent_team'
+    ]:
+        if col in df.columns:
+            df[col] = df[col].replace(repl)
+    ## if game if is also in the DF, replace it
+    if 'game_id' in df.columns:
+        ## id col ##
+        df['game_id'] = (
+            df['season'].astype('str') + '_' +
+            df['week'].astype('str').str.zfill(2) + '_' +
+            df['away_team'] + '_' +
+            df['home_team']
+        )
+    return df

nfelodcm-0.1.9/nfelodcm/Engine/Assignments/assignments/score_clean.py

@@ -0,0 +1,47 @@
+import pandas as pd
+import numpy
+
+manual_clean_dict = {
+    "2009_17_IND_BUF": {
+        "home_score": 30,
+        "away_score": 7
+    },
+    "2013_07_CIN_DET": {
+        "home_score": 24,
+        "away_score": 27
+    },
+    "2015_06_ARI_PIT": {
+        "home_score": 25,
+        "away_score": 13
+    },
+    "2015_09_PHI_DAL": {
+        "home_score": 27,
+        "away_score": 33
+    },
+    "2015_15_KC_BAL": {
+        "home_score": 14,
+        "away_score": 34
+    },
+    "2016_01_MIN_TEN": {
+        "home_score": 16,
+        "away_score": 25
+    },
+    "2016_05_NE_CLE": {
+        "home_score": 13,
+        "away_score": 33
+    }
+}
+
+def score_clean(df):
+    '''
+    Manually clean scores that are incorrect in the
+    game file
+    '''
+    ## create a home and away map ##
+    home_map = {k:manual_clean_dict[k]['home_score'] for k in manual_clean_dict.keys()}
+    away_map = {k:manual_clean_dict[k]['away_score'] for k in manual_clean_dict.keys()}
+    ## apply ##
+    df['home_score'] = df['game_id'].map(home_map).combine_first(df['home_score'])
+    df['away_score'] = df['game_id'].map(away_map).combine_first(df['away_score'])
+    ## return 
+    return df

{nfelodcm-0.1.7 → nfelodcm-0.1.9}/nfelodcm/Engine/DCMTable.py

@@ -49,6 +49,12 @@ class DCMTable():
         """
         Handles loading of the table based on freshness status.
         """
+        ## add a check to always make sure there is a local file in the DB.
+        ## If a CSV is deleted or a write fails, timestamps, and therefore
+        ## freshness, may incorrectly assume we have a current file to read
+        if not self.localIO.loadable():
+            ## if the csv does not exist, flag the update flag ##
+            self.freshness.needs_update=True
         if self.freshness.freshness_check_time is not None:
             ## if freshness was checked, update the time in the config ##
             self.tableMap.config['freshness']['last_freshness_check'] = self.freshness.freshness_check_time

{nfelodcm-0.1.7 → nfelodcm-0.1.9}/nfelodcm/Engine/Primatives/Freshness.py

@@ -159,6 +159,8 @@ class Freshness():
             ## handle response ##
             if r.status_code != 200:
                 print('WARN: Request to github commits failed.')
+                print('      Returned {0}'.format(r.status_code))
+                print('      Msg: {0}'.format(r.content))
                 return
             r = r.json()
             if len(r) == 0:

{nfelodcm-0.1.7 → nfelodcm-0.1.9}/nfelodcm/Engine/Primatives/LocalIO.py

@@ -1,4 +1,5 @@
 import pandas as pd
+import os
 import numpy
 import json
 
@@ -28,10 +29,46 @@ class LocalIO():
                     self.columns.append(col[0])
                     self.dtypes[col[0]] = col[1]
       
+    def local_exists(self):
+        """
+        Checks to make sure their is a local CSV at the filepath
+        """
+        return os.path.exists(self.data_location)
+
+    def cols_pass(self):
+        """
+        Checks the local file for columns and matches to 
+        cols int eh config
+        """
+        ## parse config cols ##
+        config_cols = list(self.config['map'].keys())
+        ## parse local cols ##
+        local_cols = []
+        if self.local_exists():
+            ## if local file exists, load the first line ##
+            local_cols = pd.read_csv(self.data_location, nrows=1).columns.tolist()
+        ## make comparison ##
+        for col in config_cols:
+            if col not in local_cols:
+                return False
+        ## if all passed, return true ##
+        return True
+
+
+    def loadable(self):
+        """
+        Runs multiple checks to make sure the file can be read
+        by the config
+        """
+        return self.local_exists() and self.cols_pass()
+
     def read(self):
         """
         Read the data from local CSV
         """
+        ## check that path exists first ##
+        if not self.loadable():
+            raise Exception('IO ERROR: Attempting read a local CSV that does not exist')
         self.df = pd.read_csv(
             self.data_location,
             usecols=self.columns,

nfelodcm-0.1.9/nfelodcm/Maps/coaches.json

@@ -0,0 +1,59 @@
+{
+  "name": "coaches",
+  "description": "Aggregated head coaching stats",
+  "last_local_update": "2024-07-07T23:13:39.454179",
+  "download_url": "https://raw.githubusercontent.com/greerreNFL/coaches/main/coaches.csv",
+  "compression": null,
+  "engine": "c",
+  "freshness": {
+    "type": "gh_commit",
+    "gh_api_endpoint": "https://api.github.com/repos/greerreNFL/coaches/commits?path=coaches.csv",
+    "gh_release_tag": null,
+    "sla_seconds": 900,
+    "last_freshness_check": "2024-07-10T17:16:06.701316+00:00"
+  },
+  "iter": {
+    "type": null,
+    "start": null
+  },
+  "assignments": [],
+  "map": {
+    "coach": "object",
+    "seasons": "int32",
+    "is_active": "int32",
+    "games": "float32",
+    "wins": "float32",
+    "losses": "float32",
+    "ties": "int32",
+    "playoff_births": "float32",
+    "games_playoff": "float32",
+    "wins_playoff": "float32",
+    "losses_playoff": "float32",
+    "ties_playoff": "float32",
+    "games_superbowl": "float32",
+    "wins_superbowl": "float32",
+    "ats_pct": "float32",
+    "ats_return": "float32",
+    "ats_risked": "float32",
+    "avg_pf": "float32",
+    "avg_pa": "float32",
+    "avg_margin": "float32",
+    "avg_spread": "float32",
+    "ats_pct_home": "float32",
+    "ats_pct_away": "float32",
+    "ats_pct_playoff": "float32",
+    "ats_pct_favorite": "float32",
+    "ats_pct_underdog": "float32",
+    "ats_pct_div": "float32",
+    "ats_pct_non_div": "float32",
+    "ats_pct_bye": "float32",
+    "ats_pct_dome": "float32",
+    "ats_roi": "float32",
+    "win_pct": "float32",
+    "win_pct_playoff": "float32",
+    "teams": "object",
+    "pfr_coach_image_url": "object",
+    "pfr_coach_tree_hired_by": "object",
+    "pfr_coach_tree_hired": "object"    
+  }
+}

nfelodcm-0.1.9/nfelodcm/Maps/filmmargins.json

@@ -0,0 +1,34 @@
+{
+  "name": "filmmargins",
+  "description": "Margins derived from PFF grades",
+  "last_local_update": "2024-07-07T23:35:11.485380",
+  "download_url": "https://raw.githubusercontent.com/greerreNFL/filmmargin/main/film_margins.csv",
+  "compression": null,
+  "engine": "c",
+  "freshness": {
+    "type": "gh_commit",
+    "gh_api_endpoint": "https://api.github.com/repos/greerreNFL/filmmargin/commits?path=film_margins.csv",
+    "gh_release_tag": null,
+    "sla_seconds": 900,
+    "last_freshness_check": "2024-07-10T17:15:32.031946+00:00"
+  },
+  "iter": {
+    "type": null,
+    "start": null
+  },
+  "assignments": [],
+  "map": {
+    "Unnamed: 0": "int32",
+    "game_id": "object",
+    "season": "int32",
+    "week": "int32",
+    "team": "object",
+    "opponent": "object",
+    "pf": "int32",
+    "pa": "int32",
+    "margin": "int32",
+    "film_margin": "float32",
+    "film_margin_predictive": "float32",
+    "film_margin_old_model": "float32"
+  }
+}

{nfelodcm-0.1.7 → nfelodcm-0.1.9}/nfelodcm/Maps/games.json

@@ -1,7 +1,7 @@
 {
   "name": "games",
   "description": "nflgamedata games",
-  "last_local_update": "2023-12-31T04:51:21.399721",
+  "last_local_update": "2024-07-07T23:13:58.469798",
   "download_url": "https://raw.githubusercontent.com/nflverse/nfldata/master/data/games.csv",
   "compression": null,
   "engine": "c",
@@ -10,14 +10,14 @@
     "gh_api_endpoint": "https://api.github.com/repos/nflverse/nfldata/commits",
     "gh_release_tag": null,
     "sla_seconds": 500,
-    "last_freshness_check": "2023-12-31T05:27:06.725310+00:00"
+    "last_freshness_check": "2024-07-10T17:15:31.355055+00:00"
   },
   "iter": {
     "type": null,
     "start": null
   },
   "assignments": [
-    "game_id_repl"
+    "fastr_team_id_repl", "score_clean"
   ],
   "map": {
     "game_id": "object",

nfelodcm-0.1.9/nfelodcm/Maps/hfa.json

@@ -0,0 +1,39 @@
+{
+  "name": "hfa",
+  "description": "Estimated home field advantage by game",
+  "last_local_update": null,
+  "download_url": "https://raw.githubusercontent.com/greerreNFL/nfelohfa/main/estimated_hfa.csv",
+  "compression": null,
+  "engine": "c",
+  "freshness": {
+    "type": "gh_commit",
+    "gh_api_endpoint": "https://api.github.com/repos/greerreNFL/nfelohfa/commits?path=estimated_hfa.csv",
+    "gh_release_tag": null,
+    "sla_seconds": 900,
+    "last_freshness_check": "2024-07-10T17:16:06.701316+00:00"
+  },
+  "iter": {
+    "type": null,
+    "start": null
+  },
+  "assignments": [],
+  "map": {
+    "game_id": "object",
+    "season": "int32",
+    "week": "int32",
+    "home_team": "object",
+    "away_team": "object",
+    "home_bye": "int32",
+    "away_bye": "int32",
+    "home_time_advantage": "float32",
+    "dif_surface": "int32",
+    "div_game": "int32",
+    "hfa_base": "float32",
+    "home_bye_adj": "float32",
+    "away_bye_adj": "float32",
+    "home_time_advantage_adj": "float32",
+    "dif_surface_adj": "float32",
+    "div_game_adj": "float32",
+    "hfa_adj": "float32"
+  }
+}

nfelodcm-0.1.9/nfelodcm/Maps/logos.json

@@ -0,0 +1,40 @@
+{
+  "name": "logos",
+  "description": "nflfastr logos",
+  "last_local_update": "2024-07-07T23:35:10.755659",
+  "download_url": "https://raw.githubusercontent.com/nflverse/nflverse-pbp/master/teams_colors_logos.csv",
+  "compression": null,
+  "engine": "c",
+  "freshness": {
+    "type": "gh_commit",
+    "gh_api_endpoint": "https://api.github.com/repos/nflverse/nflverse-pbp/commits?path=teams_colors_logos.csv",
+    "gh_release_tag": null,
+    "sla_seconds": 900,
+    "last_freshness_check": "2024-07-10T17:15:31.722504+00:00"
+  },
+  "iter": {
+    "type": null,
+    "start": null
+  },
+  "assignments": [
+    "fastr_team_id_repl"
+  ],
+  "map": {
+    "team_abbr": "object",
+    "team_name": "object",
+    "team_id": "int32",
+    "team_nick": "object",
+    "team_conf": "object",
+    "team_division": "object",
+    "team_color": "object",
+    "team_color2": "object",
+    "team_color3": "object",
+    "team_color4": "object",
+    "team_logo_wikipedia": "object",
+    "team_logo_espn": "object",
+    "team_wordmark": "object",
+    "team_conference_logo": "object",
+    "team_league_logo": "object",
+    "team_logo_squared": "object"
+  }
+}

nfelodcm-0.1.9/nfelodcm/Maps/market_data.json

@@ -0,0 +1,60 @@
+{
+  "name": "market_data",
+  "description": "Spread and totals from various sources",
+  "last_local_update": "2024-07-08T03:34:47.332681",
+  "download_url": "https://raw.githubusercontent.com/greerreNFL/nfelomarket_data/main/Data/lines.csv",
+  "compression": null,
+  "engine": "c",
+  "freshness": {
+    "type": "gh_commit",
+    "gh_api_endpoint": "https://api.github.com/repos/greerreNFL/nfelomarket_data/commits?path=Data/lines.csv",
+    "gh_release_tag": null,
+    "sla_seconds": 900,
+    "last_freshness_check": "2024-07-10T17:16:07.623265+00:00"
+  },
+  "iter": {
+    "type": null,
+    "start": null
+  },
+  "assignments": [],
+  "map": {
+    "Unnamed: 0": "int32",
+    "game_id": "object",
+    "season": "int32",
+    "week": "int32",
+    "home_team": "object",
+    "away_team": "object",
+    "home_spread_open": "float32",
+    "home_spread_open_price": "float32",
+    "away_spread_open_price": "float32",
+    "home_spread_open_source": "object",
+    "home_spread_open_timestamp": "object",
+    "home_spread_last": "float32",
+    "home_spread_last_price": "float32",
+    "away_spread_last_price": "float32",
+    "home_spread_last_source": "object",
+    "home_spread_last_timestamp": "object",
+    "home_spread_tickets_pct": "float32",
+    "home_spread_money_pct": "float32",
+    "home_spread_pcts_source": "object",
+    "home_spread_pct_timestamp": "object",
+    "home_ml_open": "float32",
+    "away_ml_open": "float32",
+    "ml_open_source": "object",
+    "ml_open_timestamp": "object",
+    "home_ml_last": "float32",
+    "away_ml_last": "float32",
+    "ml_last_source": "object",
+    "ml_last_timestamp": "object",
+    "total_line_open": "float32",
+    "under_price_open": "float32",
+    "over_price_open": "float32",
+    "total_line_open_source": "object",
+    "total_line_open_timestamp": "object",
+    "total_line_last": "float32",
+    "under_price_last": "float32",
+    "over_price_last": "float32",
+    "total_line_last_source": "object",
+    "total_line_last_timestamp": "object"
+  }
+}

{nfelodcm-0.1.7 → nfelodcm-0.1.9}/nfelodcm/Maps/participants.json

@@ -1,7 +1,7 @@
 {
   "name": "participants",
   "description": "nflfastR participants",
-  "last_local_update": null,
+  "last_local_update": "2024-07-10T17:16:25.862176",
   "download_url": "https://github.com/nflverse/nflverse-data/releases/download/pbp_participation/pbp_participation_{0}.csv",
   "compression": null,
   "engine": "c",
@@ -10,14 +10,17 @@
     "gh_api_endpoint": "https://api.github.com/repos/nflverse/nflverse-data/releases/tags",
     "gh_release_tag": "pbp_participation",
     "sla_seconds": 900,
-    "last_freshness_check": null
+    "last_freshness_check": "2024-07-10T17:16:08.727121+00:00"
   },
   "iter": {
     "type": "season",
     "start": 2016
   },
-  "assignments": [],
+  "assignments": [
+    "fastr_team_id_repl"
+  ],
   "map": {
+    "nflverse_game_id": "object",
     "old_game_id": "int32",
     "play_id": "int32",
     "possession_team": "object",
@@ -35,7 +38,7 @@
     "time_to_throw": "float32",
     "was_pressure": "object",
     "route": "object",
-    "defense_man_zone_type": "float32",
-    "defense_coverage_type": "float32"
+    "defense_man_zone_type": "object",
+    "defense_coverage_type": "object"
   }
 }

{nfelodcm-0.1.7 → nfelodcm-0.1.9}/nfelodcm/Maps/pbp.json

@@ -1,7 +1,7 @@
 {
   "name": "pbp",
   "description": "nflfastR play-by-play",
-  "last_local_update": "2023-12-28T05:48:27.222659",
+  "last_local_update": "2024-07-07T23:37:14.767747",
   "download_url": "https://github.com/nflverse/nflverse-data/releases/download/pbp/play_by_play_{0}.csv.gz",
   "compression": "gzip",
   "engine": "c",
@@ -10,14 +10,16 @@
     "gh_api_endpoint": "https://api.github.com/repos/nflverse/nflverse-data/releases/tags",
     "gh_release_tag": "pbp",
     "sla_seconds": 900,
-    "last_freshness_check": "2023-12-28T05:46:15.982851+00:00"
+    "last_freshness_check": "2024-07-10T17:15:32.328227+00:00"
   },
   "iter": {
     "type": "season",
     "start": 1999
   },
   "assignments": [
-    "game_id_repl", "penalty_formatting", "desc_based_plays"
+    "fastr_team_id_repl",
+    "penalty_formatting",
+    "desc_based_plays"
   ],
   "map": {
     "play_id": "int32",

nfelodcm-0.1.9/nfelodcm/Maps/player_stats.json

@@ -0,0 +1,77 @@
+{
+  "name": "player_stats",
+  "description": "nflfastR player stats",
+  "last_local_update": "2024-07-07T23:37:18.361644",
+  "download_url": "https://github.com/nflverse/nflverse-data/releases/download/player_stats/player_stats.csv.gz",
+  "compression": "gzip",
+  "engine": "c",
+  "freshness": {
+    "type": "gh_release",
+    "gh_api_endpoint": "https://api.github.com/repos/nflverse/nflverse-data/releases/tags",
+    "gh_release_tag": "player_stats",
+    "sla_seconds": 900,
+    "last_freshness_check": "2024-07-10T17:16:05.954026+00:00"
+  },
+  "iter": {
+    "type": null,
+    "start": null
+  },
+  "assignments": [
+    "fastr_team_id_repl"
+  ],
+  "map": {
+    "player_id": "object",
+    "player_name": "object",
+    "player_display_name": "object",
+    "position": "object",
+    "position_group": "object",
+    "headshot_url": "object",
+    "recent_team": "object",
+    "season": "int32",
+    "week": "int32",
+    "season_type": "object",
+    "opponent_team": "object",
+    "completions": "int32",
+    "attempts": "int32",
+    "passing_yards": "int32",
+    "passing_tds": "int32",
+    "interceptions": "int32",
+    "sacks": "int32",
+    "sack_yards": "int32",
+    "sack_fumbles": "int32",
+    "sack_fumbles_lost": "int32",
+    "passing_air_yards": "int32",
+    "passing_yards_after_catch": "int32",
+    "passing_first_downs": "int32",
+    "passing_epa": "float32",
+    "passing_2pt_conversions": "int32",
+    "pacr": "float32",
+    "dakota": "float32",
+    "carries": "int32",
+    "rushing_yards": "int32",
+    "rushing_tds": "int32",
+    "rushing_fumbles": "int32",
+    "rushing_fumbles_lost": "int32",
+    "rushing_first_downs": "int32",
+    "rushing_epa": "float32",
+    "rushing_2pt_conversions": "int32",
+    "receptions": "int32",
+    "targets": "int32",
+    "receiving_yards": "int32",
+    "receiving_tds": "int32",
+    "receiving_fumbles": "int32",
+    "receiving_fumbles_lost": "int32",
+    "receiving_air_yards": "int32",
+    "receiving_yards_after_catch": "int32",
+    "receiving_first_downs": "int32",
+    "receiving_epa": "float32",
+    "receiving_2pt_conversions": "int32",
+    "racr": "float32",
+    "target_share": "float32",
+    "air_yards_share": "float32",
+    "wopr": "float32",
+    "special_teams_tds": "int32",
+    "fantasy_points": "float32",
+    "fantasy_points_ppr": "float32"
+  }
+}