Hello,

I hope these set of videos are helpful:

https://www.youtube.com/watch?time_continue=609&v=ZBw2NFQ9H_g

https://www.youtube.com/watch?v=uO9Q827q2rA

This is part of the foundations course of Machine Learning_

https://courses.edx.org/courses/course-v1:ColumbiaX+DS102X+1T2017/course/

Using k-fold cross validation validation is a good way to remove bias. What k-fold does is it folds the dataset and takes various (random in default) portions of it to train the model, and then test it on the remaining instances, which is a pretty good way to check how good the model works for different train and test samples. The convention is to use 10-fold and average their results (accuracy, f1 score, precision, loss etc.) which is more reliable.

You can find more about k-fold here:

https://machinelearningmastery.com/k-fold-cross-validation/

https://www.researchgate.net/post/Why_there_is_a_need_for_K-fold_cross_validation_regardless_if_you_have_validation_set_in_the_dataset