使用 pd.crosstab 按日期降序排序

Question

I am trying to download data from the NYC COVID website, do some cleaning, and then do a crosstab with dates as columns and the region as rows.

The problem is that when I do the crosstab. the dates are presented in ascending order from left to right where I would like to see the most recent date on the left and then get older as the columns move to the right.

I'd really appreciate some help - this has me stumped. Thanks!

So my headers currently look like:|

week_ending	2020-08-08	2020-08-15

And I need them to look like:|

week_ending	2021-02-13	2021-02-06

The code follows

import pandas as pd

trendszipcode='https://raw.githubusercontent.com/nychealth/coronavirus-data/master/trends/percentpositive-by-modzcta.csv'


trendszipcode = pd.read_csv(trendszipcode,delimiter=",", index_col="week_ending").reset_index()

df=pd.melt(trendszipcode, id_vars=['week_ending'])
df[['Toss','Region']] = df['variable'].str.split('_',expand=True)
df[['Region','week_ending','value']]
df.week_ending=pd.to_datetime(df.week_ending)
df_historical=df[['Region','week_ending','value']]
df_historical=df_historical.sort_values(by='week_ending', ascending=False)
pd.crosstab(df_historical.Region, df_historical.week_ending, values=df_historical.value, aggfunc='sum').round(2)

Answer 1

You can use DataFrame.sort_index with ascending=False and axis=1 for descending sorting columns fileld by DatetimeIndex :

(pd.crosstab(df_historical.Region,df_historical.week_ending,values=df_historical.value,aggfunc='sum')
   .round(2)
   .sort_index(axis=1, ascending=False))