Q1) Should highly correlated features with the target variable be included or removed from classification and regression problems? Is there a better/elegant explanation to this step?
Actually there's no strong reason either to keep or remove features which have a low correlation with the target response, other than reducing the number of features if ...
for feature engineering there are different methods.
Pearson Correlation comes under Filter methods. Filter methods gives intuition on the high level. This can be the first step for feature engineering. In this process
the features having high correlation with target should be considered.
the features having high correlation among themselves should also ...
I would recommend Tamara Broderick's 3-part tutorial series from MLSS. She explains both modeling + inference using these methods from the ground up and focuses on the intuition.
Links to part 1, part 2, and part 3.
Non-parametric machine learning algorithms try to make assumptions about the data given the patterns observed from similar instances. By not making assumptions, they are free to learn any functional form from the training data and hence are flexible.
Unlike parametric approach, where the number of parameters are fixed, in non-parametric approaches the number ...
For theory Tibshirani: The elements of statistical learning
Also Andrew NG and other books from deeplearning.ai:
Machine Learning Yearning
of course the applied machine learning books on computer languages:
An introduction into ...
Non-parametric machine learning algorithms try to make assumptions about the data given the patterns observed from similar instances.By not making assumptions, they are free to learn any functional form from the training data.
Unlike parametric approach, where the number of parameters are fixed,
in non-parametric approaches the number of parameters grow ...
No - non-parametric methods only means that the method does not assume a function form of the data. There are non-parametric methods such as Random Forest that do not always overfit. In fact nonparametric methods could underfit, it could lack the ability to fit the training data. An example of this would be a decision stump.