5
$\begingroup$

I am new to data science and machine learning and looking for some help. I am trying to train a machine with following data set:

enter image description here

Here, L3 is the target variable. As it can be seen that the target variable contains a set of possible values, I am wondering how I can train the machine with such a data. In my test data, I might want to predict any of the one or more values given one value. Example test case :

20 c aa

'aa' is given value and I want to predict other values based on the values 20, c and 'aa'. Is it possible to do what I am trying to achieve?

Thank you.

$\endgroup$
2
$\begingroup$

Because your test cases may have values from L3 as input, this strikes me as more of a recommender system; consider encoding each possible value in L3 as a new binary column (thinking of them as products that each row/user has purchased / expressed interest in), and look into different kinds of recommender systems to see what seems most appropriate for your data.

(I'm not too familiar with recommenders. AIUI, these are all clustering algorithms, but with different notions of distance and focusing on either the products or the users. sklearn does not appear to have any built-ins specifically for recommendations, but you could use it for an underlying model. There are also some other Python packages specifically for recommenders, but I can't recommend any [see what I did there?].)

$\endgroup$
0
1
$\begingroup$

It is possible.

It is called Multi-output regression. Allows you to predict more that one variable:

Multioutput regression

$\endgroup$
1

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.