import datetime
last['date'] = last['date'].apply(lambda x:datetime.datetime.strptime(str(x),'%Y%m%d').strftime('%Y/%m/%d'))

cu.rename(columns={"Unnamed: 0":"date"},inplace=True)

traindata=pd.DataFrame(traindata,dtype=np.float)

list 转化Data Frame

a = [['a', '1.2', '4.2'], ['b', '70', '0.03'], ['x', '5', '0']]  

df = pd.DataFrame(a, columns=['one', 'two', 'three'])