32bytes/data_insight.py at main · MLFinalProject/32bytes · 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
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from datetime import date
import math

from dataset import Dataset

class DataInsight():
    def __init__(self):
        self.dataset = Dataset()
        self.train_df = self.dataset.train_test_df[self.dataset.train_test_df['dataset'].eq(self.dataset.data_label["train"])].drop(['dataset'], axis=1)
        self.train_df['adr'] = self.dataset.target_df['adr']
        self.train_df['is_canceled'] = self.dataset.target_df['is_canceled']

    def plot_attribute_adr(self, attribute):
        attribute_element = self.dataset.get_train_column(attribute).to_numpy().squeeze()
        # print(attribute)
        adr = self.dataset.get_train_adr().to_numpy().squeeze()
        # print(adr)
        plt.figure()
        plt.plot(attribute_element, adr, ',')
        plt.savefig("./data_adr_analysis/img/{}_adr.png".format(attribute))
        # plt.show()

    def plot_attribute_adr_mean(self, attribute):
        attribute_element = self.dataset.get_train_column(attribute).to_numpy().squeeze()
        attribute_element = list(set(attribute_element))

        if attribute == "arrival_date_day_of_month":
            attribute_element = [int(i) for i in attribute_element]
            attribute_element.sort()
            attribute_element = [str(i) for i in attribute_element]
            # print(attribute_element)
            # print(type(attribute_element[0]))
        else:
            attribute_element.sort()

        avg_list = []
        num_list = []
        std_list = []

        for i in attribute_element:
            avg = np.average(self.train_df.loc[self.train_df[attribute] == i].loc[:, 'adr'].to_numpy())
            std = np.std(self.train_df.loc[self.train_df[attribute] == i].loc[:, 'adr'].to_numpy())
            num = np.average(len(self.train_df.loc[self.train_df[attribute] == i].loc[:, 'adr'].to_numpy()))
            avg_list.append(avg)
            num_list.append(num)
            std_list.append(std)
            # print("average of {} = {}, which has {} samples".format(i, avg, num))
        # print(len(avg_list))
        # print(len(std_list))
        plt.figure()
        plt.errorbar(attribute_element, avg_list, std_list, linestyle='None', marker='^')
        plt.xlabel(attribute)
        plt.ylabel('adr_w_std')
        # plt.savefig("./data_adr_analysis/img/{}_adr_mean.png".format(attribute))
        plt.savefig("./data_adr_analysis/img_w_std/{}_adr_mean_std.png".format(attribute))
        # plt.show()

        # plt.figure()
        # plt.plot(attribute_element, num_list, 'b.')
        # plt.xlabel(attribute)
        # plt.ylabel('num')
        # # plt.savefig("./data_num_analysis/img/{}_num.png".format(attribute))
        # plt.show()
        # # plt.show()

        data_num_df = pd.DataFrame({attribute: num_list})
        data_num_df.to_csv(r'./data_num_analysis/csv/{}_num.csv'.format(attribute), index=True)

    def plot_attribute_is_canceled_mean(self, attribute):
        attribute_element = self.dataset.get_train_column(attribute).to_numpy().squeeze()
        attribute_element = list(set(attribute_element))
        attribute_element.sort()
        avg_list = []
        num_list = []
        for i in attribute_element:
            avg = np.average(self.train_df.loc[self.train_df[attribute] == i].loc[:, 'is_canceled'].to_numpy())
            num = np.average(len(self.train_df.loc[self.train_df[attribute] == i].loc[:, 'is_canceled'].to_numpy()))
            avg_list.append(avg)
            num_list.append(num)
            # print("average of {} = {}, which has {} samples".format(i, avg, num))
        plt.figure()
        plt.plot(attribute_element, avg_list, 'r.')
        plt.savefig("./data_is_canceled_analysis/img/{}_is_canceled_mean.png".format(attribute))

    def compare_train_test(self, attribute):
        train_attribute = self.dataset.get_train_column(attribute).to_numpy().squeeze()
        test_attribute = self.dataset.get_test_column(attribute).to_numpy().squeeze()
        train_attribute_set, test_attribute_set = set(train_attribute), set(test_attribute)

        train_unique_set = train_attribute_set - test_attribute_set
        return train_unique_set

    def plot_train_unique(self, attribute):
        train_unique_set = self.compare_train_test(attribute)
        train_unique_list = list(train_unique_set)
        train_unique_list.sort()

        avg_is_canceled_list = []
        avg_adr_list = []
        num_list = []

        for i in train_unique_list:
            avg_is_canceled = np.average(self.train_df.loc[self.train_df[attribute] == i].loc[:, 'is_canceled'].to_numpy())
            avg_adr = np.average(self.train_df.loc[self.train_df[attribute] == i].loc[:, 'adr'].to_numpy())
            num = np.average(len(self.train_df.loc[self.train_df[attribute] == i].loc[:, 'is_canceled'].to_numpy()))
            avg_is_canceled_list.append(avg_is_canceled)
            avg_adr_list.append(avg_adr)
            num_list.append(num)

        # plot is canceled
        plt.figure()
        plt.plot(train_unique_list, avg_is_canceled_list, 'r.')
        plt.savefig("./train_unique_is_canceled_analysis/{}_is_canceled_mean.png".format(attribute))

        # plot adr
        plt.figure()
        plt.plot(train_unique_list, avg_adr_list, 'r.')
        plt.savefig("./train_unique_adr_analysis/{}_adr_mean.png".format(attribute))

        # plot num
        plt.figure()
        plt.plot(train_unique_list, num_list, 'r.')
        plt.savefig("./train_unique_adr_analysis/{}_num.png".format(attribute))

    def plot_test(self, attribute):
        test_attribute_list = list(set(self.dataset.get_test_column(attribute).to_numpy().squeeze()))
        test_attribute_list.sort()

        avg_is_canceled_list = []
        avg_adr_list = []
        num_list = []

        for i in test_attribute_list:
            avg_is_canceled = np.average(self.train_df.loc[self.train_df[attribute] == i].loc[:, 'is_canceled'].to_numpy())
            avg_adr = np.average(self.train_df.loc[self.train_df[attribute] == i].loc[:, 'adr'].to_numpy())
            num = np.average(len(self.train_df.loc[self.train_df[attribute] == i].loc[:, 'is_canceled'].to_numpy()))
            avg_is_canceled_list.append(avg_is_canceled)
            avg_adr_list.append(avg_adr)
            num_list.append(num)

        # plot is canceled
        plt.figure()
        plt.plot(test_attribute_list, avg_is_canceled_list, 'r.')
        plt.savefig("./test_is_canceled_analysis/{}_is_canceled_mean.png".format(attribute))

        # plot adr
        plt.figure()
        plt.plot(test_attribute_list, avg_adr_list, 'r.')
        plt.savefig("./test_adr_analysis/{}_adr_mean.png".format(attribute))

        # plot num
        plt.figure()
        plt.plot(test_attribute_list, num_list, 'r.')
        plt.savefig("./test_adr_analysis/{}_num.png".format(attribute))