汇总 pandas 中列的观察结果

Question

假设我有一个大 Dataframe DS_df，其中包含列名 year、dealamount 和 CCS。对于从 1985 年到 2020 年的每一年，我都需要一个单独的熊猫系列，即 sum_2019。我需要总结交易金额，如果 CCS 确实发生多次（如果只发生一次，则应该将其添加到系列中）并且年份匹配：

    year    dealamount  CCS
0   2013    37,522,700  Albania_Azerbaijan
1   2013    37,522,700  Albania_Azerbaijan
2   2016    436,341,300 Albania_Greece
3   2019    763,189,200 Albania_Russia
4   2019    763,189,200 Albania_Russia
5   2019    763,189,200 Albania_Russia
6   2019    763,189,200 Albania_Russia
7   2017    150,931,000 Albania_Turkey
8   2016    275,293,750 Albania_Turkey
9   2009    258,328,000 Albania_Turkey
10  2019    153,452,000 Albania_Venezuela
11  2019    153,452,000 Albania_Venezuela
11  2017    153,452,000 Albania_Venezuela

所以在这种情况下，sum_2019 应该是一个熊猫系列，索引是 CCS，总交易量是“观察”。

Albania_Russia 3,052,756,800
Albania_Venezuela 306,904

同样，sum_2013：

Albania_Azerbaijan 75,045,400

非常感谢任何帮助，因为我需要很多数据点并且感觉很迷茫（python 真的很新）我将如何正确地自动化它？

谢谢！！

Answer 1

你想要这个吗？

df.dealamount = df.dealamount.str.replace(',','').astype(int)
new_df = df.groupby(['year','CCS']).agg({'dealamount': sum})

输出-

                         dealamount
year CCS                           
2009 Albania_Turkey       258328000
2013 Albania_Azerbaijan    75045400
2016 Albania_Greece       436341300
     Albania_Turkey       275293750
2017 Albania_Turkey       150931000
     Albania_Venezuela    153452000
2019 Albania_Russia      3052756800
     Albania_Venezuela    306904000

Answer 2

# to avoid scientific notation (e notation)
pd.set_option('display.float_format', lambda x: '%.d' % x) 

# first filter by 'year' then group by 'CSS' and finally sum by 'dealamount'
sum_2019 = df[df['year']==2019].groupby('CCS')['dealamount'].sum()

print(sum_2019)
CCS
Albania_Russia      3052756800
Albania_Venezuela    306904000
Name: dealamount, dtype: float64

汇总 pandas 中列的观察结果

Summing observations from column in pandas

python

series

multiple-columns

dataframe

pandas