Imports ..

Functions

Notes

New Fitness Scaling Method
Selection Methods
Is Crossover Necessary?

Software versions ..

Probs / To Do

Read In the Raw Data

Read No Noise Full File PSD Data

And noisy data .. (Jay says it's 'Green' channel)

Read the Solution (Alphas)

For fun let's look at the alphas (The 'Solution')**

### Data Prep ..

PCA Intro and Reversal

Make fitness cases by stitching together ins and out

Scale everything

Do the PCAs on both nonoise and noisy data

PCA's from SolveIt

Testing

Set up the variables

Scale the outs

Compute the input column means

These are required to get the alphas back into real space

Solve for Alphas

Make sure $R^2$ is good ..

Get the Alphas into Real Space

Didn't work with reduced columns ... even though $R^2$ was excellent on the fitted alphas and the cumulative variance was excellent for two columns with the PCA.

Now with EC Generated PCA Alphas

Set up the variables

Convert the ECPolys to ECAlphas

Note: Here is how to do an (inverse_transform)[https://stackoverflow.com/questions/32750915/pca-inverse-transform-manually)

Compute the polynomial

Note that the length of the fitness case inputs is 20, and the range of x's is
$\ \ \ \ \ 0.0 < x < 3.0$

Scale the outs

Compute the input column means

These are required to get the alphas back into real space

Get the Alphas into Real Space

Save this plot style ..