6 votes
Accepted

What can be done with same samples with different target?

This is completely normal; leave them in. An easy example is in an ANOVA problem (which can be viewed as a regression) where multiple subjects in the same group (so same group "value" where ...
user avatar
  • 2,830
3 votes

How to use LAT/LNG as predictor variables

If you want to predict at locations where you don't have data, and you assume that there is a continuous surface of your variable of interest (ie it is defined at all locations) then you can use ...
user avatar
  • 1,932
2 votes
Accepted

Combine multiple duplicate categorical variables into a single one for multiple linear regression

The issue is just that you consider the list of actors as ordered, but if they are considered as an (unordered) set it works perfectly. The regular "bag of words" representation used in text ...
user avatar
  • 21.8k
2 votes
Accepted

How is uncertainty evaluated for results obtained via machine learning techniques?

Here are some of the main approaches I'm aware of. One method is to use Bayesian machine learning, which learns a probability distribution over the entire parameter space (see Joris Baan's A ...
user avatar
  • 252
1 vote
Accepted

How does Catboost regressor deal with categorical features at predict time?

In a simplified way of putting it, we substitute the category id with the mean value of the training set target for this category. CatBoost implements some tricks like only using the preceding values ...
user avatar
  • 156
1 vote

What can be done with same samples with different target?

As others have explained for regression problems, leave them in. For categorical, I think the situation is a bit more nuanced. It could be that you are actually in a multi-label setting, or it could ...
user avatar
1 vote

What can be done with same samples with different target?

in addition to the answer from @Dave, consider that usually we start an assumption on the conditional distribution $P(y|x)$, for example for regression we assume Gaussian noise... this means that the ...
user avatar
1 vote

MAE divided by median metric

You are encoding a Laplace prior over your targets... now, by itself a loss has no much meaning, however, if you associate it with a distribution, you can understand how good it is the Mean Absolute ...
user avatar
1 vote

How is uncertainty evaluated for results obtained via machine learning techniques?

I am also not aware of any papers that 'prove' that ML, in general, works. Many techniques are based on optimization, distance measures, that really do not have any probability counterparts. I am also ...
user avatar
1 vote
Accepted

Loss function to prevent estimator bias

Thank you @Nikos M. for your suggestions. I was about to use your post-applied factor but then gave it another try. And found what caused this. It was that the final layer was using a ...
user avatar

Only top scored, non community-wiki answers of a minimum length are eligible