Got Changing the dtype of the features after fit() is not supported exception while predict.

[shabir1]

Got Changing the dtype of the features after fit() is not supported exception while predict.

I am getting an exception while prediction, Changing the dtype of the features after fit() is not supported.
E.g If my feature F1 has a value of 10 during the fit and while predicting my F1 value is 10.1 or vice versa then I am getting the exception.

raise ValueError("Changing the dtype of the features after fit() is "                                                                                                                                          
ValueError: Changing the dtype of the features after fit() is not supported. Fit() method was called with ['Int64', 'Int64', 'Float64', 'Float64', 'Int64', 'Int64', 'Int64', 'Float64', 'Float64', 'Float64'] wher
eas the new features have ['Int64', 'Int64', 'Int64', 'Float64', 'Int64', 'Int64', 'Int64', 'Float64', 'Float64', 'Int64'] as type 

How can I solve this issue?

[ravinkohli]

Actually, the dtype is calculated for the whole column. So, if the column was Float64 when fitting and Int64 while predicting, this error is raised. The best way to get rid of this error is by explicitly setting the dtypes for the columns that are now Int64 to Float64 using data = data.astype({'col1': 'float64', 'col2': 'float64'}) where the data is the test set and col1 and col2 are the columns which have different feature types