astel
  • Member for 2 years, 9 months
  • Last seen this week
2 answers
11 votes
14k views
Validation vs. test vs. training accuracy. Which one should I compare for claiming overfit?
6 votes

Cross validation splits your data into K folds. Each fold contains a set of training data and test data. You are correct that you get K different error rates that you then take the mean of. These ...

View answer
2 answers
5 votes
264 views
Why is 10 considered the default value for k-fold cross-validation?
4 votes

The statement: We most often use k=10 because evidence shows it's the best value for k. Smaller values don't give as good estimates, and larger values don't provide much better results either. Is ...

View answer
2 answers
2 votes
129 views
Feature Importance without Random Forest Feature Importances
3 votes

There are many ways to try to estimate feature importance. Personally I think the random forest measures get overused simply due to the fact that they have “importance” in their name and many people ...

View answer
2 answers
1 votes
23 views
Is Cross Validation needed for regression if you already know the predictors in your model?
Accepted answer
1 votes

Ask yourself why you perform cross validation. Contrary to what Dave's answer says, the point of cross-validation is to estimate your generalization error, that is how your model will perform on ...

View answer
1 answers
0 votes
196 views
Final Model fitting - subset vs entire training data
Accepted answer
1 votes

The general machine learning process is this: Split your data into two parts, training and test. So in your example I would take 100k for test and 900k for training (don't know why you say only take ...

View answer
2 answers
3 votes
1k views
Decision Trees and Feature Selection
1 votes

A decision tree has implicit feature selection during the model building process. That is, when it is building the tree, it only does so by splitting on features that cause the greatest increase in ...

View answer
4 answers
1 votes
607 views
When to normalize or regularize features in Data Science
1 votes

Step one would be understanding how the algorithms you are using work. Certain algorithms work better (distance based usually) when scaled, others don’t (like random forest). Knowing how an algorithm ...

View answer
1 answers
1 votes
31 views
Remove rows that are too much alike not to be duplicates
1 votes

Take a look into record linkage methods. They typically are used to find the same entity between two datasets but can also be used to link a dataset to itself and find duplicate records (it’s called ...

View answer
3 answers
3 votes
2k views
why do we need row sampling in random forests?
1 votes

First I think your understanding of “column sampling” is incorrect. Random forest try’s a subset of features for each split. It does not sample without replacement within an individual tree. Random ...

View answer
2 answers
2 votes
66 views
Machine Learning Validation Set
-1 votes

Comparing models cannot (or should not) be done using a test set alone. You should always have a final set of data held out to estimate your generalization error. Let’s say you compare 100 different ...

View answer