DS_MachineLearning/4_Last_factor_analysis_selection.py at master · mariazm/DS_MachineLearning · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
import pandas as pd
import numpy as np
from sklearn import decomposition
import matplotlib.pyplot as plt
from sklearn.grid_search import GridSearchCV
from sklearn.ensemble import GradientBoostingClassifier
from sklearn.ensemble  import RandomForestClassifier
from sklearn.linear_model import LogisticRegression
from sklearn.tree import DecisionTreeClassifier
from sklearn.cross_validation import KFold
from sklearn.metrics import  precision_score, recall_score,f1_score, brier_score_loss,roc_curve, auc, accuracy_score
from sklearn.calibration import  calibration_curve


def fit_model(X_train, k ):
    '''
    This function calculates returns a factor object
    '''
    fa  = decomposition.FactorAnalysis(n_components= k, copy= True, random_state=1)
    fa.fit(X_train)
    return fa

def create_factor_loads(X_train, k ):
    '''
    This function calculates the betas of the factor analysis, the weights
    '''
    fa =  fit_model(X_train, k)
    #Factor loads are the "betas" of the linear combination
    data_loads= pd.DataFrame(fa.components_)
    return data_loads

def create_factor_score(X_train, X_test, k ):
    '''
    This function returns the transformation of X_train and X_test datasets to k factors which will
    be used later to modeling
    '''
    fa =  fit_model(X_train, k)
    fa_tr = pd.DataFrame(fa.transform(X_train))
    fa_te = pd.DataFrame(fa.transform(X_test))
    return fa_tr, fa_te


def cross_validation(x_train,y_train, n_folds=5 ):
    # DB received is the training set

    models = {'logreg': LogisticRegression(), 'dectree': DecisionTreeClassifier(), 'rf':RandomForestClassifier()}

    params = {'logreg':{'C':[10**i for i in range(-2, 2)], 'penalty':['l1', 'l2']},
              'dectree':{'min_samples_leaf':[50, 100, 500, 1000], 'criterion':['entropy']} ,
              'rf':{'n_estimators':[100, 200, 500, 1000], 'criterion':['entropy']}}

    #1. Split the and prepare data
    kfolds = KFold(x_train.shape[0], n_folds = n_folds)

    best_models = {}
    for classifier in models.keys():

        best_models[classifier] = GridSearchCV(models[classifier], params[classifier], cv = kfolds, scoring = 'roc_auc')
        best_models[classifier].fit(x_train, y_train)

    return best_models


def define_best_models_k(X_train, X_test,  Y_train, k, n_folds =5):
    fa_train, fa_test = create_factor_score(X_train, X_test, k)
    best_models =  cross_validation(fa_train, Y_train, n_folds)
    return best_models, fa_test , fa_train


def k_loop(X_train, X_test, Y_train, Y_test, list_k, n_folds=5):
    results = []
    models = []
    for k in list_k:
        best_models, fa_test, fa_train = define_best_models_k(X_train, X_test, Y_train, k, n_folds)
        for model in best_models.keys() :
            params  =  best_models[model].best_params_
            auc_cv =  best_models[model].best_score_
            #Test
            pred = best_models[model].predict_proba(fa_test)[:,1]
            fpr, tpr, thresholds = roc_curve(Y_test, pred)
            roc_auc = auc(fpr, tpr)
            acc =  accuracy_score(Y_test, best_models[model].predict(fa_test))

            #Train
            pred1 = best_models[model].predict_proba(fa_train)[:,1]
            fpr1, tpr1, thresholds1 = roc_curve(Y_train, pred1)
            roc_auc1 = auc(fpr1, tpr1)
            acc1 =  accuracy_score(Y_train, best_models[model].predict(fa_train))

            results.append([k, model, roc_auc, roc_auc1, auc_cv, acc, acc1, params])
        models.append([k, best_models])
    return results, models

def get_list_per_factor(loads, X_columns):
    #Getting loads and updating col names
    var = loads
    var.columns =  X_columns

    #Get the absolute value of the load and its maximum per var
    var_abs =  pd.DataFrame(abs(var))
    idx_max =  var_abs.idxmax()

    var_load=[]
    zero_coef = []
    for i in range(var.shape[0]):
        set_idx =  pd.DataFrame(idx_max[idx_max ==i])
        set_idx.reset_index(inplace=True)
        var_load.append(list(set_idx['index']))

    return var_load, idx_max


if __name__ == '__main__':
	#Change file name and locations (Path)
	data_train = pd.read_pickle( "../Preprocessing/Factor/Datasets/Mod1/1_train_acp12_cat")
	data_test = pd.read_pickle( "../Preprocessing/Factor/Datasets/Mod1/1_test_acp12_cat")

	#In the algorithm only X must be inputted if there are more, select them
	X_train =  data_train[data_train.columns[2:]]
	Y_train = data_train["target"]

	X_test =  data_test[data_test.columns[2:]]
	Y_test = data_test["target"]

   	k_list1 =  [10,13, 16, 150, 200, 500]
   	results_1, models_1 =  k_loop(X_train, X_test, Y_train, Y_test,k_list1 , n_folds=5)
   	print pd.DataFrame(results_1, columns = ["K", "BestModel", "AUC_test", "AUC_train", "AUC_CV", "Acc_test", "Acc_train", "params"])

	##Generating input for Mary
	loads_1 = create_factor_loads(X_train, 16)
	var_load1, idx_max1  = get_list_per_factor(loads_1, X_train.columns)
	var_load1 = pd.DataFrame(var_load1).transpose()
	var_load1.to_csv("varlat16_max_mod1.csv")