pass and remove wrong id-CodePudding

I want to apply a function to my dataframe and remove rows/ids which give an error. I also want to avoid for loops as they are a bit slow for big dataframes. An example dataframe could look like this:

d = {'a': [1, "wrong_element"], 'b': [1, 2]}
df = pd.DataFrame(data=d, index=[1, 2])

print(df)

Output:

         a       b
1        1       1
2  wrong_element 2

    try:
        df['a'] = df['a'].apply(lambda x: x-2)
    except Exception:
        pass

desired output:

         a       b
1       -1       1

CodePudding user response：

You can set a NaN in case it fails, then drop the NaN

def operation(value):
    try:
        return value - 2
    except:
        return np.nan

df = pd.DataFrame({'a': [1, "wrong_element"], 'b': [0, 2]}, [1, 2])
df['a'] = df['a'].apply(operation)
df = df.dropna()

     a  b
1 -1.0  0

CodePudding user response：

The solution can be also be expressed as:

df = pd.DataFrame({'a': [1, "wrong_element"], 'b': [0, 2]}, [1, 2])

df["a"] = df["a"].replace("wrong_element", np.nan).sub(2)
df.dropna()