How can ı change nan or string values to average of the column to which it belongs?-CodePudding

df = pd.read_csv(self.table_name)

for j in df.values:
    for k in j[0:-1]:
        try:
            k = float(k)
        except ValueError:
            df.replace(to_replace=k,value=np.nan,inplace=True)

df.replace(to_replace=np.nan, value=df.mean(), inplace=True)



# df.fillna(df.mean(), inplace=True)
df.to_csv(self.table_name, index=False)
print(df)

If the data has string values while entering the training, it may not enter the training. In order to prevent this, I created a function that becomes active with a button, but the string values are deleted in the first run, and in the second run, I get the result I want. I made the button and its function over pyqt5. When the user clicks the button I mentioned, he connects to this function and its functions respectively. But where is the problem I could not solve, is there anyone who can help?

CodePudding user response：

You can use custom function for all columns without first selected by DataFrame.iloc with convert values to numeric with to_numeric and errors='coerce', so if created missing values for not parseable values. Last replace them by mean in Series.fillna:

def f(x):
    s = pd.to_numeric(x, errors='coerce')
    return s.fillna(s.mean())


df.iloc[:, 1:]= df.iloc[:, 1:].apply(f)