I have a pandas data frame column where I have float values and string type NA value. I need to replace these NAs with the mean using the following code.
trainTestJoin["col1"] = trainTestJoin.groupby("col2")["col1"].
transform(lambda x: x.fillna(x.median()))
I am getting
TypeError: could not convert string to float: NA
I tried to convert to before filling it.
trainTestJoin["LotFrontage"].astype(float)
But it gives the same issue. How to solve this issue?
Convert to numeric using pd.to_numeric which supports, via errors='coerce', conversion to float NaN for non-convertible values:
df['col1'] = pd.to_numeric(df['col1'], errors='coerce')
Then use groupby + transform directly:
df['col1'] = df['col1'].fillna(df.groupby('col2')['col1'].transform('mean'))
Or simply replace:
trainTestJoin['col1'] = trainTestJoin['col1'].replace('NA',np.nan)
And then simply:
trainTestJoin['col1'] = trainTestJoin['col1'].fillna(trainTestJoin.groupby('col2')['col1'].transform('mean'))
And now (after all):
print(trainTestJoin)
Is gonna be expected output.
