Overview

This repository contains 3 python files in the code folder: adaline.py, adalineimplementation.py, randomdata.py. adaline.py contains the actual adaline model. randomdata.py is a random data generator that is used in adalineimplementation.py to provide a specific example of how one can implement the ADALINE model found in adaline.py.

Notes

I implemented a tanh activation function because it better captures non-linearity in data as opposed to an identity activation function. The use of the hyperbolic tangent function also caused my weights to converge much faster as compared to an identity activation function because of its boundedness:

I also included a step activation function as I found that that also worked when testing with the specifications of adalineimplementation.py, although it converged much slower as compared to the use of the hyperbolic tangent function:

Here as an example call to adalineimplementation showing what the weights converged to as well as the accuracy of the model:

Next Steps

Test on real data with many more features.