https://stats.stackexchange.com/questions/262616/roc-vs-precision-recall-curves-on-imbalanced-dataset
https://neptune.ai/blog/f1-score-accuracy-roc-auc-pr-auc
https://icaiit.org/proceedings/6th_ICAIIT/1_3Fayzrakhmanov.pdf The problem of fraudulent transactions detection, which is an imbalanced classification problem (most transactions are not fraudulent), you want to maximize both precision and recall; so the area under the PR curve. As a matter of fact, the question asks you to focus on detecting fraudulent transactions (maximize true positive rate, a.k.a. Recall) while minimizing false positives (a.k.a. maximizing Precision). Another way to see it is this: for imbalanced problems like this one you'll get a lot of true negatives even from a bad model (it's easy to guess a transaction as "non-fraudulent" because most of them are!), and with high TN the ROC curve goes high fast, which would be misleading. So you wanna avoid dealing with true negatives in your evaluation, which is precisely what the PR curve allows you to do.

If we just put inside the Training set , Validation set and Test set , randomly Text, Paragraph or sentences the model will have the ability to learn specific qualities about The Author's use of language beyond just his own articles. Therefore the model will mixed up different opinions. Rather if we divided things up a the author level, so that given authors were only on the training data, or only in the test data or only in the validation data. The model will find more difficult to get a high accuracy on the test validation (What is correct and have more sense!). Because it will need to really focus in author by author articles rather than get a single political affiliation based on a bunch of mixed articles from different authors. https://developers.google.com/machine-learning/crash-course/18th-century-literature For example, suppose you are training a model with purchase data from a number of stores. You know, however, that the model will be used primarily to make predictions for stores that are not in the training data. To ensure that the model can generalize to unseen stores, you should segregate your data sets by stores. In other words, your test set should include only stores different from the evaluation set, and the evaluation set should include only stores different from the training set. https://cloud.google.com/automl-tables/docs/prepare#ml-use

Explanation/Reference: https://aws.amazon.com/step-functions/use-cases/