1. Make a KNN program that can use data with an arbitrary number of features (1 or more).

2. (To test this, you can use sklearn.datasets.make_blobs(n_features=3) to generate test data with 3 features, for example.)

3. Make your KNN program able to classify an arbitrary number of classes (2 or more).

4. Adapt your KNN program to use Manhattan distance. Are the results different? If so, what is different? Advanced: Do the same with Mahalanobis distance. Note: you will need to calculate the covariance matrix (you can use np.cov(X_train, rowvar=False)).

Please answer all these - as per standard practice, you can answer 4 as because all are related. Please answer all or ignore. Do not provide partial or wrong / copy answers.

Please don't provide the wrong java solution that is available in Chegg.

Answer in Python Code - use basic packages - Matlab, pyplot , Panda, NumPy and etc

Thanks in advance.