在给定特定条件的情况下，如何在 pandas dataframe 上使用 .count()？

Question

I have tried both of these:

    df["Buys"] = df.groupby(['symbol','month','counterparty','side'])[df[df['side']=='Buy']].count()
    df["Sells"] = df.groupby(['symbol','month','counterparty','side'])[df['side']=='Sell'].count()

But for the first I get the error "DataFrame is not hashable" and for the second "True is not a key". What am I doing wrong here?

Answer 1

Okay, I think I half figured it out. I did

dfBuys = df.groupby(['symbol','month'])['side'].apply(lambda x: (x=='Buy').sum()).reset_index(name='count')
df["Buys"] = dfBuys["count"]

dfBuys["count"] is exactly what I'm looking for, but for whatever reason sometimes df["Buys"] doesn't populate, but for the most part it works.