第7章【拓展思考】航空公司客户流失分析

"""
进行了数据清洗，属性规约，数据变换等基本预处理操作
"""
import pandas as pd


def clean(fileName):
    '''
    数据清洗，去除空记录
    :param fileName:
    :return:
    '''
    cleanedFile = './extra/data_cleaned.csv'
    data = pd.read_excel(fileName)
    # 非空保留
    data = data[data['SUM_YR_1'].notnull() & data['SUM_YR_2'].notnull()]
    # 只保留票价非零的，或者平均折扣率与总飞行公里数同时为0的记录。
    index1 = data['SUM_YR_1'] != 0
    index2 = data['SUM_YR_2'] != 0
    index3 = (data['SEG_KM_SUM'] == 0) & (data['avg_discount'] == 0)
    data = data[index1 | index2 | index3]
    data.to_csv(cleanedFile)


def change(fileName):
    '''
    取出需要的属性列
    :param fileName:
    :return:
    '''
    changedFile = './extra/data_changed.csv'
    data = pd.read_csv(fileName, encoding='utf-8')
    data = data[['LOAD_TIME', 'FFP_DATE', 'LAST_TO_END', 'FLIGHT_COUNT', 'avg_discount', 'SEG_KM_SUM', 'LAST_TO_END',
                 'P1Y_Flight_Count', 'L1Y_Flight_Count']]
    data.to_csv(changedFile, encoding='utf-8')


def LRFMCK(fileName):
    '''
    经过计算得到我的指标数据
    :param fileName:
    :return:
    '''
    data = pd.read_csv(fileName)
    # 其中K为标签标示用户类型
    data2 = pd.DataFrame(columns=['L', 'R', 'F', 'M', 'C', 'K'])
    time_list = []
    import datetime
    for i in range(len(data['LOAD_TIME'])):
        str1 = data['LOAD_TIME'][i].split('/')
        str2 = data['FFP_DATE'][i].split('/')
        temp = datetime.datetime(int(str1[0]), int(str1[1]), int(str1[2])) - datetime.datetime(int(str2[0]), int(str2[1]), int(str2[2]))
        time_list.append(temp.days)
    data2['L'] = pd.Series(time_list)
    data2['R'] = data['LAST_TO_END']
    data2['F'] = data['FLIGHT_COUNT']
    data2['M'] = data['SEG_KM_SUM']
    data2['C'] = data['avg_discount']
    temp = data['L1Y_Flight_Count'] / data['P1Y_Flight_Count']
    for i in range(len(temp)):
        if temp[i] >=0.9:
            # 未流失客户
            temp[i] = 'A'
        elif 0.5 < temp[i] < 0.9:
            # 已经流失客户
            temp[i] = 'B'
        else:
            temp[i] = 'C'
    data2['K'] = temp
    data2.to_csv('./extra/data_changed2.csv', encoding='utf-8')


def standard():
    '''
    标准差标准化
    :return:
    '''
    data = pd.read_csv('./extra/data_changed2.csv', encoding='utf-8').iloc[:, 1:6]
    zscoredfile = './extra/data_standard.csv'
    # 简洁的语句实现了标准化变换，类似地可以实现任何想要的变换
    data = (data - data.mean(axis=0)) / (data.std(axis=0))
    data.columns = ['Z' + i for i in data.columns]
    data2 = pd.read_csv('./extra/data_changed2.csv', encoding='utf-8')
    data['K'] = data2['K']
    data.to_csv(zscoredfile, index=False)


if __name__ == '__main__':
    # dataFile = './extra/data.xls'
    # clean(dataFile)
    # change('./extra/data_cleaned.csv')
    # LRFMCK('./extra/data_changed.csv')
    standard()

import pandas as pd
from sklearn import tree
from sklearn.model_selection import train_test_split
import pydotplus
# from sklearn.externals.six import StringIO
from six import StringIO

 
 
def getDataSet(fileName):
    # 读取数据
    data = pd.read_csv(fileName)
    dataSet = []
    for item in data.values:
        dataSet.append(list(item[:5]))
    label = list(data['K'])
    return dataSet, label
 
 
def divide(dataSet, labels):
    '''
    分类数据，按比例拆开
    :param dataSet:
    :param labels:
    :return:
    '''
    train_data, test_data, train_label, test_label = train_test_split(dataSet, labels, test_size=0.2)
    return train_data, test_data, train_label, test_label