0
$\begingroup$

I ran a TF IDF algorithm and the result of predicted similarities using cosine similarity is a log-normal distribution. Is this a feature of the algorithm (e.g., all logit probabilities are log-normal), or might it be a result in my data?

Thanks.

$\endgroup$
1
$\begingroup$

Let’s look at the formula used for the calculation of tf-idf scores

enter image description here

You can see a log with the idf term. The following answer explains this pretty well. log with idf

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.