New answers tagged

1

Performance Measures for Multi-Class Problems For classification problems, classifier performance is typically defined according to the confusion matrix associated with the classifier. Based on the entries of the matrix, it is possible to compute sensitivity (recall), specificity, and precision. For a single cutoff, these quantities lead to balanced accuracy ...


2

You have a combination of two problems in your data: imbalance missing values In your experiments there's a confusion about what the true distribution of the data is (or should be): either the "real data" is 97% no, or the "real data" is after removing missing values in which case it's almost balanced. It's very important to decide this ...


0

Do I correctly understand, that the test data is the whole dataset, whereas training is only a subset of it? Training and test data must not overlap. The test is a measure of quality on unseen, unfamiliar data. In the case of inbalanced data and two class classification the naive classifier, predicting always the most probable class has the quality 891 / ...


1

If you look at the imblearn documentation for classification_report_imbalanced, you can see that iba stands for "index balanced accuracy". For more information on what the index balanced accuracy is and it's value in cases on imbalanced datasets, have a look at the original paper.


0

I think you can speak of imbalanced targets if (in case of a binary classification problem) the classes are not represented in a 50:50 manner. This is almost always the case. With about 25/75 in your case, I would see this as „imbalanced“. There are some strategies to deal with this problem, such as (re)sampling so that you achieve a 50:50 balanced sample (...


0

With complex data it's rare not to have any overfitting (or underfitting). Ideally one wants to avoid strong overfitting, and if given a choice between two models it's clearly safer to use one which isn't overfit. But if it's impossible to avoid, from a practical perspective there's no reason not to use a model for this reason, in my opinion. The problem ...


0

For binary classification you can easily adjust the decision threshold to bias the classifier on either class. By default the decision threshold is 0.5. You can easily change that to e.g. 0.55 to get more 0s predictions (i.e.. if p<0.55 then 0, else 1). Such decision threshold adjustment, also called threshold moving or bias/gain adaptation, leads to ROC ...


1

My two cents: a general way to think about this process is in terms of learning and transformations. Scaling (standardization) is a transformation that you apply to every sample both in your training and test/validation/production set. These transformations are done using parameters that are learned using the training set. The aim of up/down sampling is to ...


Top 50 recent answers are included