concatenate multiple string columns with a separator in presence of empty value in pandas-CodePudding

I have a pandas dataframe like below:

df = pd.DataFrame({'id': [1,2,3], 
             'fea1':['high', 'med', ''],
             'fea2': ['med', '', 'high'],
             'fea3': ['low', 'high', 'low']})

id  fea1    fea2    fea3
1   high    med     low
2   med             high
3           high    low

I need to create a new column fea which concatenate columns fea1, fea2, fea3 with a separator /. Ignore those empty cells.

id  fea1    fea2    fea3    fea
1   high    med     low     high/med/low
2   med             high    med/high
3           high    low     high/low

Here is what I tried:

df = pd.DataFrame({'id': [1,2,3], 
             'fea1':['high', 'med', ''],
             'fea2': ['med', '', 'high'],
             'fea3': ['low', 'high', 'low']})
df['fea'] = df['fea1'] '/'   df['fea2']  '/'  df['fea3']

id  fea1    fea2    fea3    fea
1   high    med     low     high/med/low
2   med             high    med//high
3           high    low     /high/low

In row 2 and 3, I have unwanted extra /. Does anyone know how to concatenate the columns without creating these extra separators / ? Thanks.

CodePudding user response：

This will remove the / at the start and in the middle of your column

df = pd.DataFrame({'id': [1,2,3], 
             'fea1':['high', 'med', ''],
             'fea2': ['med', '', 'high'],
             'fea3': ['low', 'high', 'low']})
df['Column'] = df[[x for x in df.columns.drop('id')]].apply(lambda x : '/'.join(x), axis = 1).str.replace('//', '/').str.strip('/')

CodePudding user response：

using apply and regex to remove leading "/"

df['fea'] = df[['fea1','fea2','fea3']].apply(lambda x: "/".join(x.str.strip()), axis=1).replace(r'^/','', regex=True)
df

    id  fea1    fea2    fea3    fea
0   1   high    med     low     high/med/low
1   2   med             high    med//high
2   3           high    low     high/low