New answers tagged classification
6
votes
Visual assessment of scatterplots acceptable?
Here is something more general. Usually a data analysis problem is driven by one or more specific research questions. Even if the aim is basically exploratory, meaning that you just want to get an ...
- 26.8k
0
votes
Should I use ROC curve in my training set after training a Random Forest classification model with k-fold cross validation?
Here are the steps that you should take:
Split dataset into training set and a test set (you could do 80/20 split, 70/30 split, etc.)
Feed the training set into the cross-validation hyperparameter ...
13
votes
Accepted
Visual assessment of scatterplots acceptable?
You ask:
to what extent is it “wrong” to merely do a visual assessment of
the data? I know that this is a simplistic way of dealing with the
data, but does this necessarily make my analysis invalid? ...
- 125k
9
votes
Visual assessment of scatterplots acceptable?
This is not a complete answer, but it should point you in a useful direction (I do not want to write "the right", because there may be more ways to skin this fish -in this case-; but it will ...
- 3,302
0
votes
ROC-AUC score in sklearn for constant predictions
In case where all predicted probabilities are equal we have only two different classifiers: one that always predict one class, and one that always predict another class.
Sklearn takes the ROC to be ...
- 19.4k
0
votes
Low Feature Importance Scores but High Precision/Recall?
feature_importances_ is normalized to sum to 1, so tells you only about relative importances; it gives you no information about the goodness of fit of the model.
(...
- 4,896
0
votes
Does artificially balancing outcomes in regression lead to poor calibration? If so, how to show the poor calibration?
I think: for regression the equivalent to the poor calibration due to arbitrary upsampling is equivalent to arbitrary upsampling regions/points with high leverage. Both will introduce arbitrary bias (...
- 1,616
0
votes
Does artificially balancing outcomes in regression lead to poor calibration? If so, how to show the poor calibration?
I started wondering if the nonlinear link function plays into how the calibration is ruined in a logistic regression. This comes from looking at plots of imbalanced data and thinking that the line ...
- 65.7k
0
votes
Binary predictive classification in R, with predictors consisting of multiple values
This is more an issue of subject-matter knowledge than statistical analysis per se. I don't know of a way to use matrices of different dimensions as predictors (perhaps someone else on this site does),...
- 97.8k
0
votes
Why sequential forward doesn't select same feature as sequential backward?
SBS and SFS got different answers because both of them sought not (i) the best features, but (ii) features that are good enough.
...
- 1
Top 50 recent answers are included