4.4

1、新增一列

trainset1['fold']  = 1 

2、导入导出CSV

trainset1=pd.read_csv("E:/anaconda3/DATA/trainset1.csv")
trainset.to_csv('C:/Users/Lenovo/Desktop/trainset.csv',index=False)

3、数据框合并（默认axis=0）

validationset=pd.concat([validationset1,validationset2,validationset3,validationset4,validationset5])

4、找到最大值的索引

#fitness是列表
bestFitnessId = np.where(fitness == np.max(fitness))
bestFitnessId = bestFitnessId[0][0]#选第一个数组的第一个元素

bestFitnessIndex = np.where(s1 == np.max(s1))[0][0]

5、.loc提取数据框某一行或某一列或特定位置元素

s1 = History.loc[:,'fitness']
History.loc[bestFitnessIndex,:][0]

6、%d

trainset=pd.read_csv("E:/anaconda3/DATA/trainset%d.csv" %a)

7、两个列表合并数据框

#array转化为list
a=list()
b=list()
c={"a的变量名" ： a,
"b的变量名" ： b}
result=pd.DataFrame(c)

8、列表ifelse条件赋值

threshold[threshold>0.5]=1
threshold[threshold<=0.5]=0

9.筛选违约样本

trainset1[trainset1['PD']==1]
trainset1[trainset1['PD']==0]