#%%
#load data
data_name = 'all_loans_proc'
LD = pd.read_csv(data_dir + data_name, parse_dates=['issue_d',])
#%%
'''Store info for each predictor as a named tuple containing the col-name within
the pandas dataframe, the full_name (human readable), and the type of normalization
to apply to that feature.'''
predictor = namedtuple('predictor', ['col_name', 'full_name', 'norm_type'])
transformer_map = {'minMax':MinMaxScaler(),
'maxAbs':MaxAbsScaler(),
'standScal':StandardScaler(),
'log_minmax': LCM.log_minmax(),
'robScal':RobustScaler()}
predictors = [
predictor('acc_now_delinq','num delinq accounts','maxAbs'),
predictor('annual_inc','annual income','log_minmax'),
predictor('collections_12_mths_ex_med','num recent collections','maxAbs'),
predictor('cr_line_dur', 'duration cred line','standScal'),
predictor('delinq_2yrs', 'num recent delinq','maxAbs'),
predictor('desc_length', 'loan desc length','maxAbs'),
predictor('dti', 'debt-income ratio','standScal'),
predictor('emp_length', 'employment length','maxAbs'),
# predictor('funded_amnt','loan amount','maxAbs'),
predictor('loan_amnt','loan amount','maxAbs'),
predictor('inq_last_6mths', 'num recent inqs','maxAbs'),
predictor('int_rate', 'interest rate','maxAbs'),
