import pandas as pd
import numpy as np
import pprint as pp
import os
import pprint as pp

from sklearn.feature_selection import SelectKBest
from sklearn.feature_selection import f_classif
from scipy.stats import kstest, ks_2samp

features_dic = {}
results_dic = {}

script_dir = os.path.dirname(__file__)
rel_path = "dane_zawaly.xlsx"
abs_file_path = os.path.join(script_dir, rel_path)

dataExcel = pd.read_excel(abs_file_path, nrows=901)
df = pd.DataFrame(dataExcel)


feature_data = df.iloc[:, :-1]
diagnose_classes = np.array(df['Klasa'])

#for feature, values in feature_data.iteritems():
    #result = ks_2samp(values, diagnose_classes)
    #result = kstest(values, 'norm')
    #results_dic[feature] = result

#ranking = sorted([(feature, result) for feature, result in results_dic.items()], key=lambda z: z[1][0], reverse=True)
#pp.pprint(ranking)

# Create an SelectKBest object to select features with two best ANOVA F-Values
fvalue_selector = SelectKBest(f_classif, k=5)

# Apply the SelectKBest object to the features and target
feature_data_kbest = fvalue_selector.fit_transform(feature_data, diagnose_classes)

print('Original number of features:', feature_data.shape[1])
print('Reduced number of features:', feature_data_kbest.shape[1])