CODEWITHSUNDEEP
pythonpandaslambdakeyerror
JuanMacD
JuanMacD

Reputation: 181

Trying to use apply to make operations over different keys in a Panda DataFrame

I have a Panda DataFrame that looks somewhat like this:

df = pd.DataFrame({'ID' : ['O60829','O60341','Q9H1R3'], 'TOTAL_COVERAGE' : ['yes','yes','no'], 'BEG_D' : ['1','1','500'], 'END_D' : ['102','25','600'], 'BEG_S' : ['1','1','1'], 'END_S': ['102','25','458']})

And I want to iter over every row, check the value of 'TOTAL_COVERAGE' and if it's 'yes', perform a mathematical operation over the other values, ie:

for index, row in df.iterrows():
    df['%']  = df.apply(lambda x : ((int(x['END_S'])*100)/int(x['END_D'])) if x['TOTAL_COVERAGE'] == 'yes' else '')

But I'm getting the error: KeyError: 'TOTAL_COVERAGE' There must be an easy fix that I'm not seeing. Thanks in advance!

Upvotes: 0

Views: 62

Answers (3)

Rob Raymond
Rob Raymond

Reputation: 31216

There's no need to do an iterrows(). Conditional logic can be done with numpy.where() to give a far more efficient solution

df = pd.DataFrame({'ID' : ['O60829','O60341','Q9H1R3'], 'TOTAL_COVERAGE' : ['yes','yes','no'], 'BEG_D' : ['1','1','500'], 'END_D' : ['102','25','600'], 'BEG_S' : ['1','1','1'], 'END_S': ['102','25','458']})
df = (df
 .assign(pct=lambda x: np.where(x["TOTAL_COVERAGE"].eq("yes"),(x['END_S'].astype(int)*100)/x['END_D'].astype(int), np.nan))
 .rename(columns={"pct":"%"})
)

output

     ID TOTAL_COVERAGE BEG_D END_D BEG_S END_S      %
 O60829            yes     1   102     1   102  100.0
 O60341            yes     1    25     1    25  100.0
 Q9H1R3             no   500   600     1   458    NaN

Upvotes: 2

Michael
Michael

Reputation: 5335

You can do it without iterrows and apply, by equating directly:

df['%'] = ''
df.loc[df['TOTAL_COVERAGE'] == 'yes', '%'] = 
    df['END_S'].astype(int) * 100 / df['END_D'].astype(int)

Upvotes: 3

akuiper
akuiper

Reputation: 215047

Your can solve it in a vectorized approach, no need for iterrows and apply:

df['%'] = (df['END_S'].astype(int) * 100 / df['END_D'].astype(int)) \
            .where(df['TOTAL_COVERAGE'] == 'yes')

df

#       ID TOTAL_COVERAGE BEG_D END_D BEG_S END_S      %
#0  O60829            yes     1   102     1   102  100.0
#1  O60341            yes     1    25     1    25  100.0
#2  Q9H1R3             no   500   600     1   458    NaN

The reason you are getting a keyError is because when you are using apply, the argument to lambda x is a column (pandas Series), which can't be used to access a specific column by it's name.

Upvotes: 2

Related Questions