I have a feature array of around 4000 elements, extracted from one source. On this array I've extracted 7 more feature from other source and now I basically have a 4007 feature array from each data point. I am trying to classify these data points based on this feature array, basically doing a 1NN with manhattan distance.

However since I am really bad at maths I'm not sure how to weigh this so my 7 elements can actually help in regards to this. It feels like those 4000 elements are way more important and on distance calculation the other 7 are insignificant.

I've also given this input to a Neural Network and i'm wondering if I need to preprocess something there aswell or can I just give it the 4007 elements to the input neurons?

I've done feature-scaling but I guess that is irrelevant to the problem I'm asking.

  • 2
    $\begingroup$ With such a high dimension on the feature space, it is worth performing some kind of dimensionality reduction, such as PCA or Non Negative Matrix Factorization... If you consider that these 7 extra features are not relevant to separate the feature space, I'd not add them $\endgroup$
    – ignatius
    May 14 '18 at 12:27

The goal of supervised machine learning is automatically learn the features weights to predict target values.

If you have target values, you can fit a machine learning algorithm (e.g., a k-nearest neighbors or a neural networks).

There is no need to pick the weights yourself, the algorithm will do it for you.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.