Frits Hermans

is there already a fix for this problem? it takes down the driver for me as well

Hi Murat, the idea of using a default string for missing values like you propose would have a negative impact on training the classifier. Two records that both have a...

Could you check if the column names are the same in the data frame that you want to predict on as the one used for training? Could you check the...

Btw, if there was something wrong with your data frame, please also let me know. Then I raise a warning with some explanation. That might be useful for future users.

I think that the size shouldn't be a problem. Most likely the selected blocking rules results in too large blocks of comparisons that are made. Could you give some more...

You could try to use blocking rules like the ones below: ``` def sorted_three_letter_abbreviation(name): words = name.lower().split() first_letters = [x[0] for x in words] return "".join(sorted(first_letters)[:3]) ``` or ``` def...

this is pretty serious as packages depending on modal cannot be installed properly now