map函数

def get_type(df_obj,col):

    tmp= df_obj[col].value_counts().to_dict()

    dic={}

    index=0

    for i in (tmp):

        dic[i]=index

        index=index+1

    return dic

 

# 每个不同的变量取什么值，由自己决定，只要是不重复的连续整数变量即可

df_obj['education_num'] = df_obj['education'].map({'illiterate': 0,'basic.4y': 1,'basic.6y': 2,'basic.9y':3,'high.school':4,

                                       'professional.course':5,'unknown':6,'university.degree':7})

df_obj['month_num'] = df_obj['month'].map({'jan': 1,'feb': 2,'mar': 3,'apr':4,'may':5,

                                       'jun':6,'jul':7,'aug':8,'sep':9,'oct':10,'nov':11,'dec':12})

df_obj['loan_num'] = df_obj['loan'].map({'no': 0,'unknown': 1,'yes': 2})

 

#请补充：如下的变量 yCounts = df_obj['marital'].value_counts().to_dict()

df_obj['marital_num'] = df_obj['marital'].map(get_type(df_obj,'marital'))

df_obj['housing_num'] = df_obj['housing'].map(get_type(df_obj,'housing'))

df_obj['contact_num'] =df_obj['contact'].map(get_type(df_obj,'contact'))

df_obj['day_of_week_num'] = df_obj['day_of_week'].map(get_type(df_obj,'day_of_week'))

df_obj['poutcome_num'] = df_obj['poutcome'].map(get_type(df_obj,'poutcome'))

catCols = ['default_num','loan_num','marital_num','housing_num','day_of_week_num','education_num','month_num','poutcome_num','y']

df_obj[catCols].head()