August 2019
Intermediate to advanced
342 pages
9h 35m
English
Similarly, as we saw in Chapter 7, Frauds Prevention with Cloud AI Solutions, when we analyze data on credit card fraud, we may face strongly unbalanced data distributions, or incorrectly classified sample datasets, which reduce the effectiveness of supervised algorithms.
We have seen how it is possible to tackle and solve the problems related to mislabeled datasets by exploiting the feedback obtained from human operators (even if this solution is often burdensome in terms of both time and specialized resources employed).
In the case of unbalanced datasets (such as credit card transactions, where the samples belonging to the class of legitimate transactions largely exceed the samples of fraud transactions), ...
Read now
Unlock full access