CODEWITHSUNDEEP
python-3.xpandas
ali bakhtiari
ali bakhtiari

Reputation: 1097

deleting specific rows of a dataframe in pandas based on the value of one row

I have a dataframe df with about 1 million rows:

    HOUSEID  PERSONID   ENDTIME STRTTIME    TRVLCMIN        
0   20000017    1        1020      955         25   
1   20000017    1        1132      1130        2    
2   20000017    1        1400      1330        30   
3   20000017    2        1020      955         25   
4   20000017    2        1025      1022        3    
5   20000017    2        -9        1120        2
6   20000017    2        2035      2000        35
7   20000231    1        952       945         7
8   20000231    1        1000      -9          5
9   20000231    2        2013      2002       -9

Each row is a trip made by a person. And each unique combination of HOUSEID and PERSONID is a person. I want to delete a person entirely if there is a "-9" in even one of the trips of these columns ENDTIME, STRTTIME and TRVLCMIN.

Here is the output of my dataframe by this explanation:

        HOUSEID PERSONID    ENDTIME    STRTTIME       TRVLCMIN      
    0   20000017    1        1020        955              25    
    1   20000017    1        1132        1130             2 
    2   20000017    1        1400        1330             30

Can you help me with that?

I very much appreciate your time.

Upvotes: 0

Views: 49

Answers (1)

Quang Hoang
Quang Hoang

Reputation: 150805

You could check for -9 in any of those columns, then groupby().transform to identify the combination HOUSEID, PERSONID that has some -9:

# rows having -9 in one of the specified columns
s = df[['ENDTIME','STRTTIME','TRVLCMIN']].eq(-9).any(1)

# combination of `HOUSEID`, `PERSONID` having some `-9`
mask = s.groupby([df['HOUSEID'],df['PERSONID']]).transform('any')

# output
df[~mask]

Output:

    HOUSEID  PERSONID  ENDTIME  STRTTIME  TRVLCMIN
0  20000017         1     1020       955        25
1  20000017         1     1132      1130         2
2  20000017         1     1400      1330        30

Upvotes: 1

Related Questions