def create_submission_file(estimator, cols, weights=None, in_path=None, out_path=None, verbose=None,
load_from_temp=None, temp_path=None):
"""
Creates the submission.txt file.
"""
# Set defaults
if verbose is None or verbose == 'WARNING':
logger.setLevel(logging.WARNING)
if verbose == 'INFO':
logger.setLevel(logging.INFO)
if verbose == 'DEBUG':
logger.setLevel(logging.INFO)
if out_path is None:
out_path = os.path.join(CONFIG.results_path, "submission.txt")
train_df, weather_df = get_df(cols, load_from_temp, temp_path)
submission_df = load_submission(in_path)
ff = FeatureFactory(submission_df.copy(), weather_df)
for col in set(cols).union({'TIME', 'WEEKEND'}):
ff(col)
sub_df = ff.X
predictions = {}
for assignment in CONFIG.submission_assignments:
logger.info('Preparing for submission: %s...' % assignment)
t_df = train_df[train_df['ASS_ASSIGNMENT'] == assignment]
X, y, _, _ = get_cross_validation_parameters(t_df, cols, weather_df=weather_df,
weights=weights)
s_df = sub_df[sub_df['ASS_ASSIGNMENT'] == assignment]
X_submission, _, _, _ = get_cross_validation_parameters(s_df, cols, weather_df=weather_df,
weights=weights, label='prediction')
estimator.fit(X, y)
predictions[assignment] = list(estimator.predict(X_submission))
sub_df['raw_prediction'] = sub_df['ASS_ASSIGNMENT'].apply(lambda x: predictions[x].pop(0))
submission_df['prediction'] = \
sub_df.apply(lambda x: max(int(x['raw_prediction'] + x['MEAN'] + 0.5), 0), axis=1)
submission_df.drop('MEAN', axis=1, inplace=True)
submission_df.to_csv(out_path, sep='\t', index=None, encoding='utf-8', date_format='%Y-%m-%d %H:%M:%S.000')
return submission_df
OrthogonalMatchingPursuit, RANSACRegressor
from sklearn.tree import DecisionTreeRegressor, ExtraTreeRegressor
from sklearn.svm import SVR, LinearSVR
from sklearn.ensemble import AdaBoostRegressor, BaggingRegressor, GradientBoostingRegressor, RandomForestRegressor
from sklearn.isotonic import IsotonicRegression
from sklearn import metrics
from sklearn.feature_selection import SelectKBest, chi2
from learning.stacked_regressor import StackedRegression
pd.options.mode.chained_assignment = None # Remove warnings
logger.setLevel(logging.DEBUG)
# _df = load_train_df(CONFIG.preprocessed_train_path)
_submission_df = load_submission()
# _estimator = KNeighborsRegressor(n_neighbors=3)
# _estimator = ARDRegression()
_estimator = BayesianRidge()
# _estimator = OrthogonalMatchingPursuit()
# _estimator = ExtraTreeRegressor()
_scoring = 'mean_squared_error'
_k_fold = 3
_n_jobs = 3
_verbose = 0
_fit_params = None
_cols = ["CUM_DAYS"]
_weights = None
_temp_path = os.path.join(os.getcwd(), 'train.csv')
preprocess_meteo3(df)
# Booleans for each department where average amount of rain is above 1mm and where it has frozen.
print('Meteo4...')
preprocess_meteo4(df)
return df
# logger.debug('Meteo dataframes concatenated.')
# logger.debug('Summing departments...')
# df = df.groupby('DATE').agg({'MIN_TEMP': lambda x: pd.Series([(x <= 2).sum()]),
# 'PRECIP': lambda x: pd.Series([(x > 1).sum()])})
# df = df.rename(columns={'MIN_TEMP': 'NUMB_FROZEN_DEPT'})
# df = df.rename(columns={'PRECIP': 'NUMB_WET_DEPT'})
# df[["NUMB_FROZEN_DEPT", 'NUMB_WET_DEPT']] = df[["NUMB_FROZEN_DEPT", "NUMB_WET_DEPT"]].astype(int)
# df.to_csv(CONFIG.preprocessed_meteo_path)
# logger.info('Saved meteo in csv file.')
if __name__ == "__main__":
# parse_train_as_dict(CONFIG.raw_train_path, CONFIG.preprocessed_train_path)
# parse_train_as_df(CONFIG.raw_train_path, CONFIG.preprocessed_train_path)
# df1 = parse_meteo_as_df(CONFIG.raw_meteo_path1)
# df2 = parse_meteo_as_df(CONFIG.raw_meteo_path2)
# df = pd.concat([df1, df2])
# print(df)
# run('train', verbose=1)
from utils import load_submission
print(complete_with_means(load_submission(CONFIG.submission_path), CONFIG.submission_path_mean))
pass
