Add other/improved method to balance class distributions

[maet3608]

https://www.analyticsvidhya.com/blog/2017/03/imbalanced-classification-problem/?utm_source=feedburner&utm_medium=email&utm_campaign=Feed%3A+AnalyticsVidhya+%28Analytics+Vidhya%29

[piyushmadan]

FAQ should be updated as the suggested method generates an error
https://maet3608.github.io/nuts-ml/faq.html?#how-to-use-class-weights-for-imbalanced-classes-in-keras

TypeError: train() takes 1 positional argument but 2 were given
class_weights = {0:1119, 1:76}
t_loss, t_acc = (ReadFromCOS('train') >> train_cache >> Shuffle(10000) >> Stratify(labelcol=1, labeldist=labeldist_train) >> build_batch >> network.train(class_weights) >> Print('Please wait, training on batches', every_sec=3000) >> NOP(Print()) >> Unzip())

instead, use
network.train(class_weight=class_weights)

[maet3608]

Thanks for reporting. Documentation has been fixed.