How are scores calculated? Also, last names with higher scores are ignored in the presence of first names.

[Otpid]

Hi, thanks for the awesome dataset. I actually have 3 questions:

1. How are the scores calculated? I read the older issue regarding this but it wasn't really answered there. If possible, could you disclose the formula that you used to calculate the score of a name? Is the score normalized? If a name has 0.35 for country A and 0.35 for country B, what does that indicate? If a name has 0.35 score as a first name for country A and 0.35 score as a surname for the same country A, what does that mean?

2. is search_first_name method no longer available?

3. It seems last names are ignored in deciding the country association. For the name "anderson", the result indicates it's a male name from Brazil. Although, "anderson" has a higher max last name score (0.59 for USA) than the max first name score (0.42 for Brazil)

[philipperemy]

@Otpid thanks!

1. It's a bit involved. I'd like to give more explanation but I need to dig into it again, which I will do later. The code is here:
   https://github.com/philipperemy/name-dataset/blob/master/generation/filter_records.py

2. search_first_name was removed. Now there's only search and it retrieves both first and last names.

   name-dataset/names_dataset/nd_v3.py

   Line 68 in ce2333c

   def search(self, name: str):

   .

3. Yes, only the first_name has been used in deciding the country association (when using .describe). The last name is discarded. We can update the behavior because it might be confusing.