def get_orig_plot_data(self, col, tail=False):
# 1 vs 2 plots in the figure
pd = {}
if tail:
num = 10
tail = self.data[col].value_counts().tail(10)
pd = tail.to_dict()
else:
num = 1000
head = self.data[col].value_counts().head(10)
pd = head.to_dict()
l = []
l.append(list(pd.keys()))
l.append(list(pd.values()))
return l
def add_to_player_dict(pd, players, date):
team_stats = {'fg': 0, 'fga': 0, 'fta': 0, 'mp': 0, 'tov': 0, 'orb': 0, 'drb': 0, 'fg3a': 0}
for name in players:
stats = players[name]
for key in pd.keys():
if key=='name':
pd[key].append(name)
elif key=='date':
pd[key].append(date)
else:
if stats==None:
pd[key].append(None)
else:
if key in team_stats:
team_stats[key] += stats[key]
pd[key].append(stats[key])
return team_stats
def evaluate_models_compare_to_stacking_r_square(predictions_dict,
true_treatment_effect,
stacking_predictions):
"""
:param predictions_dict:
:param true_treatment_effect:
:param stacking_predictions: an array of predicitons given by the stacking model
:return: a table/array of evaluation metrics for each model
"""
pd = copy.deepcopy(predictions_dict)
if "Actuals" in predictions_dict:
pd.pop('Actuals')
if "generated_data" in predictions_dict:
pd.pop('generated_data')
r2_dict = {}
for key in pd.keys():
r2_dict[key] = r2_score(true_treatment_effect, pd[key])
r2_dict['stacking'] = r2_score(true_treatment_effect, stacking_predictions)
return r2_dict
def get_colums(pd):
return " (" + ", ".join(str(key) for key in pd.keys()) + ") "