使用 shift 函数比较行值

Question

我正在学习熊猫，我遇到了以下方法来比较数据帧中的行。

在这里，我使用np.were和shift()函数来比较列中的值。

import pandas as pd 
import numpy as np
# Initialise data to Dicts of series. 
d = {'col' : pd.Series([10, 30, 20, 40, 70, 60])} 
# creates Dataframe. 
df = pd.DataFrame(d)
df['Relation'] = np.where(df['col'] > df['col'].shift(), "Grater", "Less")
df

此处的输出如下所示：

col Relation
0   10  Less
1   30  Grater
2   20  Less
3   40  Grater
4   70  Grater
5   60  Less

我在第 3 行感到困惑，为什么它显示为Grater？，40 小于 70，所以它应该显示为Less.我在这里做错了什么？

Answer 1

因为将40与20进行比较，因为移位索引按1：

df['Relation'] = np.where(df['col'] > df['col'].shift(), "Grater", "Less")
df['shifted'] = df['col'].shift()
df['m'] = df['col'] > df['col'].shift()
print (df)
col Relation  shifted      m
0   10     Less      NaN  False
1   30   Grater     10.0   True
2   20     Less     30.0  False
3   40   Grater     20.0   True <- here
4   70   Grater     40.0   True
5   60     Less     70.0  False

也许你想按-1换档：

df['Relation'] = np.where(df['col'] > df['col'].shift(-1), "Grater", "Less")
df['shifted'] = df['col'].shift(-1)
df['m'] = df['col'] > df['col'].shift(-1)
print (df)
col Relation  shifted      m
0   10     Less     30.0  False
1   30   Grater     20.0   True
2   20     Less     40.0  False
3   40     Less     70.0  False
4   70   Grater     60.0   True
5   60     Less      NaN  False