taufiqbashori

# define dependent and independent variables
# un-transformed dataset
xi = coded_df[[_i for _i in coded_df_vif_iter2.variables]]
yi = coded_df.WORK_LIFE_BALANCE_SCORE

# transformed dataset
xt = coded_df_T[[_i for _i in coded_df_T_vif_iter2.variables]]
yt = coded_df_T.WORK_LIFE_BALANCE_SCORE

taufiqbashori

#coded_df_T_vif_iter2.sort_values(by='VIF', ascending = False).head(12)

from matplotlib import pyplot as plt
plt.rcParams["figure.figsize"] = [14, 7]
plt.rcParams["figure.autolayout"] = True
f, axes = plt.subplots(1, 2)
# UnTransformed Data
VIF_plot = sns.barplot(x=coded_df_vif_iter2.sort_values(by='VIF', ascending = False)["variables"],
                       y=coded_df_vif_iter2.sort_values(by='VIF', ascending = False)["VIF"], ax=axes[0]) 
for item in VIF_plot.get_xticklabels():

taufiqbashori

# dropping high VIF variables one by one
# calculate VIF
coded_df_vif_iter2 = calc_vif(coded_df.drop(columns = ["WORK_LIFE_BALANCE_SCORE","SLEEP_HOURS","TODO_COMPLETED","SOCIAL_NETWORK","FRUITS_VEGGIES",
                                                       "SUPPORTING_OTHERS","PERSONAL_AWARDS","WEEKLY_MEDITATION"]))
coded_df_T_vif_iter2 = calc_vif(coded_df_T.drop(columns = ["WORK_LIFE_BALANCE_SCORE","SLEEP_HOURS_T","ACHIEVEMENT_T","SOCIAL_NETWORK","FRUITS_VEGGIES",
                                                           "FLOW_T","TODO_COMPLETED_T","SUPPORTING_OTHERS","WEEKLY_MEDITATION"]))

coded_df_vif_iter2.sort_values(by='VIF', ascending = False).head(12)

taufiqbashori

# Import library for VIF
from statsmodels.stats.outliers_influence import variance_inflation_factor

def calc_vif(X):

    # Calculating VIF
    vif = pd.DataFrame()
    vif["variables"] = X.columns
    vif["VIF"] = [variance_inflation_factor(X.values, i) for i in range(X.shape[1])]

taufiqbashori

# transformed dataset and non transformed dataset
# non transformed dataset
coded_df.head()
# transformed dataset
coded_df_T = pd.concat((coded_df.drop(columns=[_i for _i in xval[0:8]]),transformed_skew_df), axis = 1)
coded_df_T.head()

taufiqbashori

skewness_dict = {}
for _i in coded_df.columns:
  if coded_df[_i].dtypes == 'int64' or coded_df[_i].dtypes == 'float64':
    skewness_dict[_i] = coded_df[_i].skew()
  else:
    None

# sort based on highest absolute skew
skewness = sorted(skewness_dict.items(), key=lambda kv: abs(kv[1]), reverse = True)

taufiqbashori

coded_df = pd.concat(
                      [df.drop(["GENDER", "AGE", "BMI_RANGE","SUFFICIENT_INCOME"],axis=1), 
                       pd.concat([transformed[_i] for _i in list_transformed], axis = 1)], 
                     axis = 1)
coded_df.head()

taufiqbashori

# dummy variables
df['BMI_RANGE'].replace({1: "BMI < 25", 2: "BMI >= 25"}, inplace = True)
df['SUFFICIENT_INCOME'].replace({1: "INCOME INSUFFICIENT", 2: "INCOME SUFFICIENT"}, inplace = True)
df['AGE'].replace(
    {'Less than 20': '0. Less than 20', 
     '21 to 35': '1. 21 to 35',
     '36 to 50': '2. 36 to 50', 
     '51 or more': '3. 51 or more',
    }, inplace = True
    )

taufiqbashori

pd.crosstab(
    df['GENDER'], 
    df['AGE'], 
    rownames=['Gender'], 
    colnames=['Age'],
    margins = True,
    normalize = True
).style.format('{:.2%}')

taufiqbashori

pd.crosstab(
    df['Month'], 
    df['Year'], 
    rownames=['Month'], 
    colnames=['Year'],
    margins = True,
    normalize = True
).style.format('{:.2%}')