search

tomerm / MLClassification

Classification using ML approach for English / Hebrew / Arabic data sets
1 stars 2 forks source link

Issues to fix #35

Open matanzuckerman opened 5 years ago

matanzuckerman commented 5 years ago

Hi @tomerm @semion1956

I noticed a few issues.

  1. Under function "data loader" in data.py if the parameter "w2vload" =false (line 50) then all the lines after it will not run. In this case I can't do tokanization for example because it's written below it (the elf.jar is not set (line 68) ) 2.The output of crossvalidation Is the F1 score of each iteration. I want to see in the end the average of all the measures from all the iterations. 3.The parameters Rankthreshold,diffthreshold in the script metrics (line 40) are hard coded. please move them to the config file Thanks
semion1956 commented 5 years ago

@matanzuckerman

  1. Problem is already solved in the latest version of the code.
  2. OK.
  3. diffthreshold isn't used. Rankthreshold - OK, but I do not recommend to change it in a large range.
matanzuckerman commented 5 years ago

@semion1956 Thanks