def impute_data(training_data: DataFrame,
testing_data: DataFrame) -> Tuple[DataFrame, DataFrame]:
"""Calculate imputations from the training data and apply to the testing data
Inputs
training_data: DataFrame
training data
testing_data: DataFrame
testing data
Returns
training_data: imputations based on the training data only
testing_data: imputations based on the training data only
"""
# calculate from training data, impute into testing data
# make training imputations
training_data.CODE_GENDER = training_data.CODE_GENDER.fillna(
get_mode(training_data.CODE_GENDER.dropna()))
training_data.NAME_TYPE_SUITE = training_data.NAME_TYPE_SUITE.fillna(
get_mode(training_data.NAME_TYPE_SUITE.dropna()))
training_data.OBS_30_CNT_SOCIAL_CIRCLE = training_data.OBS_30_CNT_SOCIAL_CIRCLE.fillna(
np.median(training_data.OBS_30_CNT_SOCIAL_CIRCLE.dropna()))
training_data.DEF_30_CNT_SOCIAL_CIRCLE = training_data.DEF_30_CNT_SOCIAL_CIRCLE.fillna(
np.median(training_data.DEF_30_CNT_SOCIAL_CIRCLE.dropna()))
training_data.OBS_60_CNT_SOCIAL_CIRCLE = training_data.OBS_60_CNT_SOCIAL_CIRCLE.fillna(
np.median(training_data.OBS_60_CNT_SOCIAL_CIRCLE.dropna()))
training_data.DEF_60_CNT_SOCIAL_CIRCLE = training_data.DEF_60_CNT_SOCIAL_CIRCLE.fillna(
np.median(training_data.DEF_60_CNT_SOCIAL_CIRCLE.dropna()))
training_data.AMT_ANNUITY = training_data.AMT_ANNUITY.fillna(
np.median(training_data.AMT_ANNUITY.dropna()))
training_data.ANNUITY_INCOME_RATIO = training_data.ANNUITY_INCOME_RATIO.fillna(
np.median(training_data.ANNUITY_INCOME_RATIO.dropna()))
training_data.DAYS_LAST_PHONE_CHANGE = training_data.DAYS_LAST_PHONE_CHANGE.fillna(
np.median(training_data.DAYS_LAST_PHONE_CHANGE.dropna()))
# make testing imputation based on training data
testing_data.CODE_GENDER = testing_data.CODE_GENDER.fillna(
get_mode(training_data.CODE_GENDER.dropna()))
testing_data.NAME_TYPE_SUITE = testing_data.NAME_TYPE_SUITE.fillna(
get_mode(training_data.NAME_TYPE_SUITE.dropna()))
testing_data.OBS_30_CNT_SOCIAL_CIRCLE = testing_data.OBS_30_CNT_SOCIAL_CIRCLE.fillna(
np.median(training_data.OBS_30_CNT_SOCIAL_CIRCLE.dropna()))
testing_data.DEF_30_CNT_SOCIAL_CIRCLE = testing_data.DEF_30_CNT_SOCIAL_CIRCLE.fillna(
np.median(training_data.DEF_30_CNT_SOCIAL_CIRCLE.dropna()))
testing_data.OBS_60_CNT_SOCIAL_CIRCLE = testing_data.OBS_60_CNT_SOCIAL_CIRCLE.fillna(
np.median(training_data.OBS_60_CNT_SOCIAL_CIRCLE.dropna()))
testing_data.DEF_60_CNT_SOCIAL_CIRCLE = testing_data.DEF_60_CNT_SOCIAL_CIRCLE.fillna(
np.median(training_data.DEF_60_CNT_SOCIAL_CIRCLE.dropna()))
testing_data.AMT_ANNUITY = testing_data.AMT_ANNUITY.fillna(
np.median(training_data.AMT_ANNUITY.dropna()))
testing_data.ANNUITY_INCOME_RATIO = testing_data.ANNUITY_INCOME_RATIO.fillna(
np.median(training_data.ANNUITY_INCOME_RATIO.dropna()))
testing_data.DAYS_LAST_PHONE_CHANGE = testing_data.DAYS_LAST_PHONE_CHANGE.fillna(
np.median(training_data.DAYS_LAST_PHONE_CHANGE.dropna()))
return training_data, testing_data
