How to get n previous rows in pandas, after using loc?-CodePudding

We have data representing workers billing history of payments and penalties after their work shifts. Sometimes penalty for the worker is wrong, because he had technical problems with his mobile app and in reality he attended the job. Later he gets his penalty refunded which goes with description 'balance_correction'. The goal is to show n lines (rows) in data to to find pattern for what he got the penalty. So here is the data:

d = {'balance_id': [70775,70775 ,70775,70775,70775], 'amount': [2500, 2450,-500,500,2700]
     ,'description':['payment_for_job_order_080ecd','payment_for_job_order_180eca'
                     ,'penalty_for_being_absent_at_job','balance_correction','payment_for_job_order_120ecq']}
df1 = pd.DataFrame(data=d)
df1


    balance_id amount   description
0    70775     2500       payment_for_job_order_080ecd
1    70775     2450       payment_for_job_order_180eca
2    70775     -500       penalty_for_being_absent_at_job
3    70775     500        balance_correction
4    70775     2700       payment_for_job_order_120ecq

I try this:

df1.loc[df1['description']=='balance_correction'].iloc[:-2]

and get nothing. Also using shift doesn't help. If we need 2 roes to show, the result should be

    balance_id  amount  description
1   70775     2450      payment_for_job_order_180eca
2   70775     -500      penalty_for_being_absent_at_job

What can solve the issue?

CodePudding user response：

If the index on your data frame is sequential (0, 1, 2, 3, ...), you can filter by the index:

idx = df1.loc[df1['description'] == 'balance_correction'].index
df1.loc[(idx - 2).append(idx - 1)]