Question

我有一个数据透视表，我想创建另一个具有相同格式的数据透视表，但是现在它包含了同比变化百分比。

这是一个简单的示例：

                                      total sales         
name                                          bye       hi
category                              electronics clothing
date       week_number quarter_number                     
2000-01-07 1           1                      NaN    123.0
2000-01-14 2           1                    456.0      NaN
2001-01-05 1           1                      NaN    180.0
2001-01-12 2           1                    350.0      NaN

产生以下数据透视表：

                                      total sales pchg Y/Y         
name                                          bye       hi
category                              electronics clothing
date       week_number quarter_number                     
2000-01-07 1           1                      NaN      NaN
2000-01-14 2           1                      NaN      NaN
2001-01-05 1           1                      NaN    0.463
2001-01-12 2           1                    -0.23      NaN

现在让我们说我想计算每年的变化百分比。生成的数据透视表如下所示：

my_data = {
    'date': [datetime.date(2000,1,7), datetime.date(2000,1,14),
             datetime.date(2001,1,5), datetime.date(2001,1,12),
             datetime.date(2000, 1, 7), datetime.date(2000, 1, 14),
             datetime.date(2001, 1, 5), datetime.date(2001, 1, 12),
             datetime.date(2000, 1, 7), datetime.date(2000, 1, 14),
             datetime.date(2001, 1, 5), datetime.date(2001, 1, 12),
             datetime.date(2000, 1, 7), datetime.date(2000, 1, 14),
             datetime.date(2001, 1, 5), datetime.date(2001, 1, 12)],
    'week_number': [1,2,1,2,1,2,1,2,1,2,1,2,1,2,1,2],
    'quarter_number': [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1],
    'name': ['hi','hi','hi','hi','hi','hi','hi','hi','bye','bye','bye','bye','bye','bye','bye','bye'],
    'category': ['clothing','clothing','clothing','clothing','electronics','electronics','electronics','electronics',
                 'clothing', 'clothing', 'clothing', 'clothing', 'electronics', 'electronics', 'electronics','electronics'],
    'total sales': [123,456,180,350,123,456,180,350,123,456,180,350,123,456,180,350]
}
my_df = pd.DataFrame(my_data)
my_df.pivot_table(index=['date','week_number','quarter_number'], columns=['name', 'category'])

my_df.pivot_table(index=['date','week_number','quarter_number'], columns=['name', 'category']).apply(pd.Series.pct_change)
                                      total sales     ...                
name                                          bye     ...              hi
category                                 clothing     ...     electronics
date       week_number quarter_number                 ...                
2000-01-07 1           1                      NaN     ...             NaN
2000-01-14 2           1                 2.707317     ...        2.707317
2001-01-05 1           1                -0.605263     ...       -0.605263
2001-01-12 2           1                 0.944444     ...        0.944444

请注意，在一般情况下，我们有N个名称，多年的数据和K个类别。

我在这里也提供了一个更一般的情况，以显示pct_change在默认模式下不起作用，因为它不会按年百分比变化。

import React from 'react';
import Layout from '../components/layout';
import {graphql} from 'gatsby';

export default ({pageContext: {locale, folderName}, data}) => {
  // if (typeof window !== `undefined`) {
    const fileFrontmatter = data.file.childMarkdownRemark.frontmatter;

    return (
      <Layout path="/" locale={locale}>
        <div>{fileFrontmatter.title}</div>
      </Layout>
    );
  // }
};

export const query = graphql`
  query Template($locale: String, $folderName: String) {
    file(name: { eq: $locale }, relativeDirectory: { eq: $folderName }) {
      childMarkdownRemark{
        frontmatter{
          title
        }
      }
    }
  }
`;

pct_change显然是错误的，因为它不提供Y / Y更改，而是提供第i到第i + 1行。

Answer 1

您可以使用pct_change来获得所需的结果：

pivoted = pd.pivot_table(my_df, index=['date','week_number','quarter_number'], columns=['name', 'category'])
pivoted.groupby(level='week_number').transform(pd.Series.pct_change)
#                                      total sales          
#name                                          bye        hi
#category                              electronics  clothing
#date       week_number quarter_number                      
#2000-01-07 1           1                      NaN       NaN
#2000-01-14 2           1                      NaN       NaN
#2001-01-05 1           1                      NaN  0.463415
#2001-01-12 2           1                -0.232456       NaN

数据透视表中Y与Y的变化

1 个答案: