如何使用 pandas groupby.count() 作为条件

Question

I have a dataframe df with two columns, Ticker and Trade Results.

I want to create a new dataframe, with three columns - Ticker, Number of Trades, Profitable Trades.

I have used the groupby and count function to get the Number of Trades column, this works fine.

My problem is with the third column, Profitable Trades, where the Trade Result is > 0, I have not found a way to get in this condition.

Creating DF (works fine)

df = pd.DataFrame(
    {'Ticker': ['[BTC]','[ETH]','[LTC]','[BTC]','[ETH]',
              '[LTC]','[BTC]','[ETH]','[LTC]'],
     'Trade Results': [5,10,5,-5,-10,-5,5,10,5]}
)

Ticker	Trade Results
BTC	5
ETH	10
LTC	5
BTC	-5
ETH	-10
LTC	-5
BTC	5
ETH	10
LTC	5

Grouping Tickers and Getting Count (works fine)

df_Grouped = df.groupby(['Ticker']).count()

Ticker	Count
BTC	3
ETH	3
LTC	3

Conditional Column (my problem)

This is the part I haven't been able to figure out, my latest attempt is below but returns NaN for the profitable column.

df_Grouped['Profitable'] = df.groupby(['Trade Result'] > 0).count()

Desired Output

Ticker	Count	Profitable
BTC	3	2
ETH	3	2
LTC	3	2

Answer 1

You can do it like this:

df_Grouped = df.groupby(['Ticker']).agg({'Trade Results': [('Count', 'count'), ('Profitable', lambda x: len(x[x>0]))]}).reset_index()

Output:

                 Count Profitable
0  BTC             3          2
1  ETH             3          2
2  LTC             3          2

Answer 2

You can always pre-filter, however I like @David Ms answer

df_Grouped['Profitable'] = df[df['Trade Results'] > 0].groupby(['Ticker']).count()

Answer 3

You could create a boolean for rows that are greater than 0, before aggregating on the groupby:

(
    df.assign(gt_0=df["Trade Results"].gt(0))
    .groupby("Ticker")
    .agg(Count=("gt_0", "size"), Profitable=("gt_0", "sum"))
)

    Count   Profitable
Ticker      
BTC     3   2
ETH     3   2
LTC     3   2

Answer 4

You can also create a new df with groupBy and then merge it.

df_Grouped = df.groupby(['Ticker']).count()
df_Grouped.reset_index(level = 'Ticker')
df_new = df[df['Trade Results'] >0].groupby(['Ticker']).count().reset_index(level = 'Ticker')
print(pd.merge(df_Grouped, df_new, left_on='Ticker', right_on='Ticker', how='left'))