CODEWITHSUNDEEP
pythonpandas
Eric
Eric

Reputation: 43

Using pandas dtype as a condition

I am trying to remove certain characters from strings within certain columns in a pandas dataframe. I am doing all of this within a for loop so I would like to use an if statement within the loop to perform the actions on all 'object' dtype columns.

for col in pitchtype :
pitchtype[col] = pitchtype[col].replace(np.nan,0) 
if pitchtype[col].dtype == 'object':
    pitchtype[col] = pitchtype[col].map(lambda x: x.replace(' %',''))

if there a way to make that condition in the if statement?

edit: added my DF below. Basically the columns with % in the header have '%' symbols in the values which are preventing them from being float. I am trying to remove the '%'s and change the columns to type float afterwards.

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 264 entries, 0 to 263

Data columns (total 18 columns):

Name        264 non-null object
Team        264 non-null object
FB%         264 non-null object
FBv         264 non-null float64
SL%         264 non-null object
SLv         264 non-null float64
CT%         264 non-null object
CTv         264 non-null float64
CB%         264 non-null object
CBv         264 non-null float64
CH%         264 non-null object
CHv         264 non-null float64
SF%         264 non-null object
SFv         264 non-null float64
KN%         264 non-null object
KNv         264 non-null float64
XX%         264 non-null object
playerid    264 non-null int64

dtypes: float64(7), int64(1), object(10)
memory usage: 37.2+ KB

Upvotes: 4

Views: 11980

Answers (2)

jpp
jpp

Reputation: 164773

You can use pd.DataFrame.select_dtypes and pd.Series.str.rstrip:

for col in df.select_dtypes(['object']):
    df[col] = pd.to_numeric(df[col].str.rstrip('%'), errors='coerce')

The conversion to float is performed by pd.to_numeric. errors='coerce' gives NaN for non-convertible values.

Upvotes: 4

Rocky Li
Rocky Li

Reputation: 5958

I think this maybe what you're looking for, checking individual object to see if they're string.

if pitchtype[col].dtype == object: # No quotes around it!
    pitchtype[col] = pitchtype[col].map(lambda x: x.replace(' %','') if type(x) == str else x)

Upvotes: 5

Related Questions