def modify_cf_df(self, bat, bowl):
if bat is None or bowl is None or bat.empty or bowl.empty:
raise RuntimeError(self.EX_MODIFY_CF.format(bat, bowl))
# bat = self.__drop_na(bat)
# bowl = self.__drop_na(bowl)
bat = self.__replace_zero(bat)
bowl = self.__replace_zero(bowl)
bat = bat.fillna(0)
bowl = bowl.fillna(0)
bat[self.TEXT_HS] = DataUtils.remove_asterisk_all(bat[self.TEXT_HS])
bowl[self.TEXT_WICKET_HAUL] = self.__wicket_haul_col(bowl)
if self.TEXT_SPAN in bat.columns:
bat = DataUtils.drop_cols(bat, self.CONSISTENCY_BAT_DROP_COLS)
bowl = DataUtils.drop_cols(bowl, self.CONSISTENCY_BOWL_DROP_COLS)
else:
bat = DataUtils.drop_cols(bat, self.FORM_BAT_DROP_COLS)
bowl = DataUtils.drop_cols(bowl, self.FORM_BOWL_DROP_COLS)
df = pd.merge(bat, bowl, on=self.TEXT_PLAYER)
return df.reset_index(drop=True)
def __form(self, soup, playing_eleven):
players = DataUtils.get_players(soup)
if len(players) > 0:
table_rows = players[0].find_all(self.TEXT_TR)
df = DataUtils.create_df(table_rows)
return DataUtils.playing_eleven_df(df, playing_eleven)
else:
return
def __consistency(self, alter_url, team, playing_eleven):
team_id = TeamsDAO.get_team_id(team)
url = alter_url.team_url(team_id)
soup = BSoup(url).get_soup()
players = DataUtils.get_players(soup)
table_rows = players[0].find_all(self.TEXT_TR)
df = DataUtils.create_df(table_rows)
return DataUtils.playing_eleven_df(df, playing_eleven)
def create_total_consistency(self, playing_eleven):
playing_eleven.sort()
total_consistency_df = pd.DataFrame()
players_id_list = PlayersDAO.get_player_id(playing_eleven)
for player_id in players_id_list:
if player_id is not None:
consistency = self.__process_tc(player_id)
else:
total_consistency_df = self.__default_values_total_consistency(
total_consistency_df)
continue
if len(consistency) > 0:
table_rows = consistency[0].find_all("tr")
df = DataUtils.create_df(table_rows)
total_consistency_df = total_consistency_df.append(
df.transpose())
else:
total_consistency_df = self.__default_values_total_consistency(
total_consistency_df)
total_consistency_df = self.__modify_tc(
total_consistency_df.reset_index(drop=True))
total_consistency_df['Player'] = playing_eleven
total_consistency_df['InnBowls'] = self.__bowl_inn_tc(
total_consistency_df)
return total_consistency_df.reset_index(drop=True)
def __modify_tc(self, df):
df = df.replace(r'^\s*$', 0, regex=True)
df = df.fillna(0)
df_bat = df.loc[:, :15]
df_bowl = df.loc[:, 16:]
fielding_row = df.loc[df[16] == 'Fielding'].index.tolist()
for row in fielding_row:
row += 1
df_bowl.loc[row] = 0
wk_row = df.loc[df[16] == 'Wicket Keeping'].index.tolist()
for row in wk_row:
row += 1
df_bowl.loc[row] = 0
df = pd.concat([df_bat, df_bowl], axis=1, ignore_index=True)
df = self.__set_columns(df)
df = DataUtils.drop_cols(df, self.TOTAL_CON_DROP_COLS)
# df = df.drop(df.columns[[7, 13, 14, 15, 16, 17]], axis=1)
# df = df.drop_duplicates(keep=False)
# df = df[~df['Innings:'].str.contains("Innings:")]
duplicates = df['Innings:'].str.contains("Innings:")
for index, value in duplicates.items():
if value is True:
df.drop(index, inplace=True)
return df
def modify_total_consistency(self, df):
if df.empty or df is None:
raise RuntimeError(self.EX_MODIFY_TC.format(df))
df = df.replace(r'^\s*$', 0, regex=True)
df = df.fillna(0)
df = self.__replace_zero(df)
df = self.__replace_na(df)
df[self.TEXT_HIGH_SCORE] = DataUtils.remove_asterisk_all(
df[self.TEXT_HIGH_SCORE])
df[self.TEXT_WICKET_HAUL] = pd.to_numeric(
df[self.TEXT_FOUR_WICKETS]) + pd.to_numeric(
df[self.TEXT_FIVE_WICKETS])
df = DataUtils.drop_cols(df, self.TOTAL_CONSISTENCY_DROP_COLS)
return df.reset_index(drop=True)
def __prepare_data(self, url, playing_eleven):
soup = BSoup(url).get_soup()
players = DataUtils.get_players(soup)
if len(players) > 0:
# Can add thread to run both simultaneously
recent_form_bat = self.__process_rf(players, 0, playing_eleven)
recent_form_bowl = self.__process_rf(players, 1, playing_eleven)
return recent_form_bat, recent_form_bowl
else:
return
def __bat_sr_venue(self, player_id_list, stadium_venue):
ser = []
for player_id in player_id_list:
total_runs = 0
total_bf = 0
if player_id is not None:
table = self.__find_table(player_id,
ConstantUrl.CAREER_INN_BAT_URL)
else:
ser.append(0)
continue
if len(table) > 0:
table_rows = table[0].find_all("tr")
df = DataUtils.create_df(table_rows)
df = df.dropna()
cols = [
'0', 'Date', 'Versus', 'Ground', 'D/N', 'How Dismissed',
'Runs', 'B/F', 'S/R', '9', 'Aggr', 'Avg', 'S/R Pro'
]
df.columns = cols
df = df.drop(['0', '9'], axis=1)
df = df.reset_index(drop=True)
for i in df.index:
if stadium_venue in df['Ground'][i]:
if 'DNB' in df['Runs'][i]:
continue
total_runs += pd.to_numeric(
DataUtils.remove_asterisk(df['Runs'][i]))
total_bf += pd.to_numeric(
DataUtils.remove_asterisk(df['B/F'][i]))
if total_runs == 0 or total_bf == 0:
ser.append(0)
else:
bat_sr = total_runs / total_bf * 100
ser.append(round(bat_sr, 2))
else:
ser.append(0)
return ser
def modify_ov_df(self, bat, bowl):
if bat is None or bowl is None or bat.empty or bowl.empty:
raise RuntimeError(self.EX_MODIFY_OV.format(bat, bowl))
bat = bat.replace(r'^\s*$', 0, regex=True)
bowl = bowl.replace(r'^\s*$', 0, regex=True)
bat = bat.fillna(0)
bowl = bowl.fillna(0)
bat[self.TEXT_HS] = DataUtils.remove_asterisk_all(bat[self.TEXT_HS])
if self.TEXT_VERSUS in bat.columns and self.TEXT_VERSUS in bowl.columns:
bat = DataUtils.drop_cols(bat, self.OPPOSITION_BAT_DROP_COLS)
bowl = DataUtils.drop_cols(bowl, self.OPPOSITION_BOWL_DROP_COLS)
elif self.TEXT_GROUND in bat.columns and self.TEXT_GROUND in bowl.columns:
bat = DataUtils.drop_cols(bat, self.VENUE_BAT_DROP_COLS)
bowl = DataUtils.drop_cols(bowl, self.VENUE_BOWL_DROP_COLS)
df = pd.merge(bat, bowl, on=self.TEXT_PLAYER)
return df.reset_index(drop=True)
def __get_zeros_df(self, player_id):
if player_id is not None:
table = self.__find_table(player_id,
ConstantUrl.CAREER_INN_BAT_URL)
else:
return
if len(table) > 0:
table_rows = table[0].find_all("tr")
df = DataUtils.create_df(table_rows)
df.columns = df.loc[2]
df = df.drop(df.index[2])
df = df.dropna()
return df
return
def __process_rf(self, players, index, playing_eleven):
table_rows = players[index].find_all(self.TEXT_TR)
df = DataUtils.create_df(table_rows)
return DataUtils.playing_eleven_df(df, playing_eleven)
def __get_df(table_rows):
df = DataUtils.create_df(table_rows)
df.columns = df.loc[0]
df = df.drop(df.index[0])
return df
def venue_range(self, venue_range):
venue_bat = {}
venue_bowl = {}
for i in self.venue.index:
venue_bat.update({
self.venue[self.COL_PLAYER][i]:
0.4262 * (DataUtils.check_range(
venue_range[1], self.venue[self.COL_BAT_AVG][i])) +
0.2566 * (DataUtils.check_range(
venue_range[0], self.venue[self.COL_NO_OF_INN_BAT][i])) +
0.1510 * (DataUtils.check_range(
venue_range[2], self.venue[self.COL_BAT_SR][i])) +
0.0787 * (DataUtils.check_range(
venue_range[3], self.venue[self.COL_CENTURIES][i])) +
0.0556 * (DataUtils.check_range(
venue_range[4], self.venue[self.COL_FIFTIES][i])) +
0.0328 * (DataUtils.check_range(venue_range[5],
self.venue[self.COL_HS][i]))
})
venue_bowl.update({
self.venue[self.COL_PLAYER][i]:
0.3018 * (DataUtils.check_range(
venue_range[6], self.venue[self.COL_OVERS][i])) +
0.2783 * (DataUtils.check_range(
venue_range[0], self.venue[self.COL_NO_OF_INN_BOWL][i])) +
0.1836 * (DataUtils.check_range(
venue_range[8], self.venue[self.COL_BOWL_SR][i])) +
0.1391 * (DataUtils.check_range(
venue_range[7], self.venue[self.COL_BOWL_AVG][i])) +
0.0972 * (DataUtils.check_range(
venue_range[9], self.venue[self.COL_WICKET_HAUL][i]))
})
return venue_bat, venue_bowl
def opposition_range(self, opposition_range):
opposition_bat = {}
opposition_bowl = {}
for i in self.opposition.index:
opposition_bat.update({
self.opposition[self.COL_PLAYER][i]:
0.4262 *
(DataUtils.check_range(opposition_range[1],
self.opposition[self.COL_BAT_AVG][i])) +
0.2566 * (DataUtils.check_range(
opposition_range[0],
self.opposition[self.COL_NO_OF_INN_BAT][i])) + 0.1510 *
(DataUtils.check_range(opposition_range[2],
self.opposition[self.COL_BAT_SR][i])) +
0.0787 *
(DataUtils.check_range(opposition_range[3],
self.opposition[self.COL_CENTURIES][i]))
+ 0.0556 *
(DataUtils.check_range(opposition_range[4],
self.opposition[self.COL_FIFTIES][i])) -
0.0328 * (DataUtils.check_range(
opposition_range[5], self.opposition[self.COL_ZEROS][i]))
})
opposition_bowl.update({
self.opposition[self.COL_PLAYER][i]:
0.3177 * (DataUtils.check_range(
opposition_range[6], self.opposition[self.COL_OVERS][i])) +
0.3177 * (DataUtils.check_range(
opposition_range[0],
self.opposition[self.COL_NO_OF_INN_BOWL][i])) + 0.1933 *
(DataUtils.check_range(opposition_range[8],
self.opposition[self.COL_BOWL_SR][i])) +
0.1465 *
(DataUtils.check_range(opposition_range[7],
self.opposition[self.COL_BOWL_AVG][i]))
+ 0.0943 * (DataUtils.check_range(
opposition_range[9],
self.opposition[self.COL_WICKET_HAUL][i]))
})
return opposition_bat, opposition_bowl
def recent_form_range(self, form_range):
recent_form_bat = {}
recent_form_bowl = {}
for i in self.recent_form.index:
recent_form_bat.update({
self.recent_form[self.COL_PLAYER][i]:
0.4262 * (DataUtils.check_range(
form_range[1], self.recent_form[self.COL_BAT_AVG][i])) +
0.2566 * (DataUtils.check_range(
form_range[0],
self.recent_form[self.COL_NO_OF_INN_BAT][i])) +
0.1510 * (DataUtils.check_range(
form_range[2], self.recent_form[self.COL_BAT_SR][i])) +
0.0787 * (DataUtils.check_range(
form_range[3], self.recent_form[self.COL_CENTURIES][i])) +
0.0556 * (DataUtils.check_range(
form_range[4], self.recent_form[self.COL_FIFTIES][i])) -
0.0328 * (DataUtils.check_range(
form_range[5], self.recent_form[self.COL_ZEROS][i]))
})
recent_form_bowl.update({
self.recent_form[self.COL_PLAYER][i]:
0.3269 * (DataUtils.check_range(
form_range[6], self.recent_form[self.COL_OVERS][i])) +
0.2846 * (DataUtils.check_range(
form_range[0],
self.recent_form[self.COL_NO_OF_INN_BOWL][i])) +
0.1877 * (DataUtils.check_range(
form_range[8], self.recent_form[self.COL_BOWL_SR][i])) +
0.1210 * (DataUtils.check_range(
form_range[7], self.recent_form[self.COL_BOWL_AVG][i])) +
0.0798 * (DataUtils.check_range(
form_range[9], self.recent_form[self.COL_WICKET_HAUL][i]))
})
return recent_form_bat, recent_form_bowl
def total_consistency_range(self, consistency_range):
total_con_bat = {}
total_con_bowl = {}
for i in self.total_consistency.index:
total_con_bat.update({
self.total_consistency[self.COL_PLAYER][i]:
0.4262 * (DataUtils.check_range(
consistency_range[1],
self.total_consistency[self.COL_BAT_AVG][i])) + 0.2566 *
(DataUtils.check_range(
consistency_range[0],
self.total_consistency[self.COL_NO_OF_INN_BAT][i])) +
0.1510 * (DataUtils.check_range(
consistency_range[2],
self.total_consistency[self.COL_BAT_SR][i])) + 0.0787 *
(DataUtils.check_range(
consistency_range[3],
self.total_consistency[self.COL_CENTURIES][i])) + 0.0556 *
(DataUtils.check_range(
consistency_range[4],
self.total_consistency[self.COL_FIFTIES][i])) - 0.0328 *
(DataUtils.check_range(
consistency_range[5],
self.total_consistency[self.COL_ZEROS][i]))
})
total_con_bowl.update({
self.total_consistency[self.COL_PLAYER][i]:
0.4174 * (DataUtils.check_range(
consistency_range[6],
self.total_consistency[self.COL_OVERS][i])) + 0.2634 *
(DataUtils.check_range(
consistency_range[0],
self.total_consistency[self.COL_NO_OF_INN_BOWL][i])) +
0.1602 * (DataUtils.check_range(
consistency_range[8],
self.total_consistency[self.COL_BOWL_SR][i])) + 0.0975 *
(DataUtils.check_range(
consistency_range[7],
self.total_consistency[self.COL_BOWL_AVG][i])) + 0.0615 *
(DataUtils.check_range(
consistency_range[9],
self.total_consistency[self.COL_WICKET_HAUL][i]))
})
return total_con_bat, total_con_bowl