Diagnosis of diabetes - neural network

Neural network made without dedicated libraries such as TensorFlow,
designed to help diagnose diabetes based on patient data.
Learning by backpropagation algorithm, explained eg. here https://towardsdatascience.com/understanding-backpropagation-abcc509ca9d0.

Visualizing data

Seaborn plot displayed at the end of the training.
We can see that the network lowers the error rate and finds the solution on its own.

Another sample plot

Here we see a different learning process
Because starting weights are random
But outcome is similiar

Console

Application also displays progress, parameters onto the console

Requirements

Pandas
Numpy
Seaborn/matplotlib