Hello everyone I'm looking to perform a binary classification (either 0 or 1) on a large sample of data (around 5 million points). The data is numerical; i.e. [1,4,6,2,3,0,3,...], and I would like to be able to train a model to accept multiple samples of pre-classified data; which would then be able to generate a prediction on an entirely new set of data. I've been looking into SVM types of learning; however it seems this may not be exactly what I want. Any suggestions or advice would be much appreciated! I'd be happy to answer any questions you may have.

This and this might be useful, especially the second one.

Perceptrons are exceedingly easy to implement and are very useful. Multi-layer perceptrons are great, especially if your data is not linearly separable.

Neural networks (depending of your algorithms) aren’t as easy to implement, but they can be even more accurate. There are many different kinds, but I recommend a feed-forward backpropagating neural network. It’s relatively simple, and should get the job done adequately.