CODEWITHSUNDEEP
pandasfor-loopconditional-statements
Hamza
Hamza

Reputation: 73

pandas : update value if condition met in loop

I have to update a data frame column if a condition is met. But there are multiple conditions and multiple values to update to. Therefore I want to do it in a loop.

The data frame is like:

mode  car1  car2  bus1  bus2
car1   10    20    5     2
car2   11    22    3     1
bus1   4     4     2     2  
bus2   3     4     3     5

I realize the data structure is slightly odd but let's go with this. If mode says car1, I want the new column value to have the value in the column car1. And so on.

My code is like:

targets = ['car1', 'car2', 'bus1', 'bus2']
for target in targets:
    df.loc[(df.mode==f'target'),'value']=df.[target]

This works but every it replaces the rows in which the condition isn't met with a NaN. Therefore, I only end up with the new value column containing the value for bus2 in bus2 rows but NaNs in all other rows.

In Stata, I would have written:

gen value = .
foreach x in car1 car2 bus1 bus2 {
replace value = `x' if mode=="`x'"
}

Looking for similar code in Python!

Upvotes: 0

Views: 956

Answers (2)

ggagliano
ggagliano

Reputation: 1074

This should work:

df['newcol'] = 0
for key, item in df.iterrows():
    df['newcol'].iloc[key] = item[item['mode']]

Upvotes: 0

BENY
BENY

Reputation: 323226

In pandas there is lookup

df['newvalue']=df.set_index('mode').lookup(df['mode'],df['mode'])
df
Out[184]: 
   mode  car1  car2  bus1  bus2  newcol  newvalue
0  car1    10    20     5     2      10        10
1  car2    11    22     3     1      22        22
2  bus1     4     4     2     2       2         2
3  bus2     3     4     3     5       5         5

Upvotes: 1

Related Questions