Native Bayes Kernel - no prediction in few examples

Serek91

Hi,

Naive Bayes Kernel operator, after finished classification, marks few predictions as "?". What can be cause of it? It can't find any similarities with rest of examples?

My csv looks like:

id, author_id, characters_number, words_number, average_sentence_length, average_word_length, ratio_unique_words, ratio_word_length_1, ratio_word_length_2, ratio_word_length_3, ratio_word_length_4, ratio_word_length_5, ratio_word_length_6, ratio_word_length_7, ratio_word_length_8, ratio_word_length_9, ratio_word_length_10, ratio_word_length_11, ratio_word_length_12, ratio_word_length_13, ratio_word_length_14, ratio_word_length_15, ratio_word_length_16, most_used_word_1, most_used_word_2, most_used_word_3, most_used_word_4
"100395", "1000866", "1640", "318", "44", "3,94", "0,6006289", "0,08805031", "0,1761006", "0,1949686", "0,2327044", "0,1069182", "0,07861635", "0,06289309", "0,02515723", "0,01257862", "0,01886792", "0,003144654", "0", "0", "0", "0", "0", "anyway", "really", "decided", "write"

rfuentealba

Hello, @Serek91.

Yes, the cause might be exactly that the algorithm is unable to find similarities for your data, or that there are nulls in certain key values that are key to detect the author_id. However, I cannot be sure if you don't send us the process and data (perhaps via private message, if your data is sensitive). I tried to replicate this and didn't get your same results.

All the best,

Rod.

Serek91

Ok, thanks. My whole csv has 500 records. There is no nulls. So I think it is because operator can't find any similarities for given set of records.