def test()->tuple:
"""
Load test dataset.
return -- tuple(dataframe data, dictionary columns)
"""
# header
click.secho('Load data..', fg='green')
# read data
df, col = csv2df(PATH_TEST, ltarget=[], lindex=['PassengerId'])
# return
return (df, col)
def train()->tuple:
"""
Load training dataset.
return -- tuple(dataframe data, dictionary columns)
"""
# header
click.secho('Load data..', fg='green')
# read data
df, col = csv2df(PATH_TRAIN, ltarget=['Survived'], lindex=['PassengerId'])
# return
return (df, col)
def load()->tuple:
"""
Load weather dataset (without target variable).
return -- tuple(dataframe data, dictionary columns)
"""
# header
click.secho('Load data..', fg='green')
# read data
ddt = {'lcol': ['dt'], 'sformat': '%Y-%m-%d %H:%M:%S'}
df, dcol = csv2df(PATH, lindex=['dt'], ddt=ddt)
# return
return (df, dcol)
def load() -> tuple:
"""
Load wine dataset (without target variable).
return -- tuple(dataframe data, dictionary columns)
"""
# header
click.secho('Load data..', fg='green')
# read data
df, dcol = csv2df(PATH)
# format
df.Proline = df.Proline.astype(float)
df.Magnesium = df.Magnesium.astype(float)
df.Alcohol = df.Alcohol.astype(int)
# update dcol
col = columns()
col.get(df)
# return
return (df, col)
def load() -> tuple:
"""
Load solar dataset.
return -- tuple(dataframe data, dictionary columns)
"""
# header
click.secho('Load data..', fg='green')
# read data
ddt = {'lcol': ['dt'], 'sformat': '%Y-%m-%d %H:%M:%S'}
df, dcol = csv2df(PATH, ltarget=['y', 'cy'], lindex=['dt'], ddt=ddt)
# format
df.cy = df.cy.astype(int)
# update dcol
col = columns()
col.get(df, ['y', 'cy'])
# return
return (df, col)
quit('Aborted!')
# fit, transform and return
return transformer.full_pipeline.fit_transform(df)
def numerical(df: 'dataframe')->'array':
"""
Launch a pre-processing Pipeline with only numerical variables.
df -- data to be transformed.
"""
# validate if there are NaN values.
if df.isnull().sum().sum() > 0:
click.secho('[error] the dataframe to be transformated contains NaN values.', fg='red', bold=True)
print(df.isnull().sum())
quit('Aborted!')
# fit, transform and return
return transformer.num_pipeline.fit_transform(df)
if __name__ == '__main__':
from tools import reader
# read data
data, dcol = reader.csv2df('../../datasets/dataset.weather.csv', lindex=['datetime'])
# get a sample
dfX = data[dcol['lc_float'][:1] + dcol['lc_cat'][:1]]
# transformation
X = full(dfX.dropna().head())
print(X[:, :5])