从分组中寻找均值并显示所有信息

Question

我有这个数据框。

df1 = pd.DataFrame({'userId': [1,1,1,2,2,3,4,4],
                   'movieId': [500,600,700,1100,1200,600,600,1900],  
                   'ratings': [3.5,4.5,2.0,5.0,4.0,4.5,5.0,3.5]})


df2 = pd.DataFrame({'userId':[1,1,2,3,4,5],
                    'movieId':[500,600,1100,800,900,600],
                    'tag':['Highly quotable','Boxing story','MMA','Tom Hardy','Fun','long movie']})


frames = [df1, df2]
result = pd.concat(frames, sort = False)
result

  userId movieId ratings tag
0   1    500     3.5    NaN
1   1   600      4.5    NaN
2   1   700      2.0    NaN
3   2   1100     5.0    NaN
4   2   1200     4.0    NaN
5   3   600      4.5    NaN
6   4   600      5.0    NaN
7   4   1900     3.5    NaN
0   1   500      NaN    Highly quotable
1   1   600      NaN    Boxing story
2   2   1100     NaN    MMA
3   3   800      NaN    Tom Hardy
4   4   900      NaN    Fun
5   5   600      NaN    long movie

我正在尝试分组 movieId。我想要的是计算每个 movie.If 计数为 2 或大于 2 的出现次数，对于这种情况，它应该取 ratings 的平均值并显示所有信息。我已经试过了，但它给出了错误。 KeyError: 'ratings'.

这是代码

group = result.groupby('movieId')['movieId'].count().reset_index(name="count")
agg = group['ratings'].mean().reset_index(name="mean")
agg
#right code here

Answer 1

我会提出一些不同的建议。我不会使用 concat，而是使用 pd.merge

看看这个：

import pandas as pd

df1 = pd.DataFrame({'userId': [1,1,1,2,2,3,4,4],
                   'movieId': [500,600,700,1100,1200,600,600,1900],
                   'ratings': [3.5,4.5,2.0,5.0,4.0,4.5,5.0,3.5]})


df2 = pd.DataFrame({'userId':[1,1,2,3,4,5],
                    'movieId':[500,600,1100,800,900,600],
                    'tag':['Highly quotable','Boxing story','MMA','Tom Hardy','Fun','long movie']})

# Merging df1 and df2, now you'll not have unnecessary NaN Values
result = df1.merge(df2[['movieId', 'tag']], on='movieId', how='left')

# Grouping by using two tipes of output with agg
result.groupby(by=['movieId', 'tag'], as_index=False).agg({'ratings': ['count', 'mean']})

输出将是：

  movieId              tag ratings          
                             count      mean
0     500  Highly quotable       1  3.500000
1     600     Boxing story       3  4.666667
2     600       long movie       3  4.666667
3    1100              MMA       1  5.000000

希望对你有用

编辑

正如你在评论中所问，如果你想过滤数据框，你可以简单地运行下面的代码：

# Removing multiindex columns (just to be easier for you)
result = result.droplevel(0, axis=1)
result.columns = ['userId', 'movieId', 'ratings_count', 'ratings_mean']

# Filtering
result = result[result['ratings_count'] >= 2]
result = result[result['ratings_mean'] >= 3]

有更好的方法来做到这一点，但我假设您还不知道如何使用 Pandas MultiIndex，所以我做了一个简单的解决方案。

从分组中寻找均值并显示所有信息

Finding mean from group by and displaying all information

python

count

mean

pandas-groupby