python - 从数据帧的段(循环)计算
问题描述
2 个数据帧。1短1长。我想将长的分成几块,使用相关系数将它们与短的进行比较。
分裂很好。然而,当把它们放在计算中时,它会返回 Nan。
import pandas as pd
data_a = {'ID': ["a1","a2","a3","a4","a5","a6","a7","a8","a9","a10","a11","a12","a13","a14","a15"],
'Unit_Weight': [178,153,193,195,214,157,205,212,219,166,217,186,170,207,204]}
df_a = pd.DataFrame(data_a)
data_b = {'ID': ["b1","b2","b3","b4","b5"],
'Unit_Weight': [128,123,123,125,204]}
df_b = pd.DataFrame(data_b)
size = 5 # 5 rows in the long data-frame
list_of_df_a = [df_a.loc[i:i+size-1,:] for i in range(0, len(df_a),size)]
for each in list_of_df_a:
corr_e = each['Unit_Weight'].corr(df_b['Unit_Weight'])
输出:
0.6797202605786716
nan
nan
出了什么问题,如何纠正?谢谢你。
ps:这些是手动计算的结果:
0.6797202605786716
-0.5501914564062937
0.2653370297540246
ID Unit_Weight
0 a1 178
1 a2 153
2 a3 193
3 a4 195
4 a5 214
ID Unit_Weight
5 a6 157
6 a7 205
7 a8 212
8 a9 219
9 a10 166
ID Unit_Weight
10 a11 217
11 a12 186
12 a13 170
13 a14 207
14 a15 204
解决方案
两者都需要相同的索引Series
,因此DataFrame.reset_index
与 一起使用drop=True
:
for each in list_of_df_a:
corr_e = each['Unit_Weight'].reset_index(drop=True).corr(df_b['Unit_Weight'])
print (corr_e)
0.6797202605786716
-0.5501914564062937
0.26533702975402457