Scree Plot for Pricipal Component Analysis

[pswaldia]

Describe the solution you'd like
I would love to have a plot that shows the amount of variance explained by the prinicipal components. It can help to reduce the dimensionality of the features to the dimensions that explains the variance in the dataset the most. It will help to find the optimal principal components without extensive searching.

Examples

This plot shows that around 350 features out of total 784 explains about 95% of the variance.

[lwgray]

@pawaldia Awesome suggestion! We are catching up on issues and PRs since we just returned from a short hiatus. We will respond properly asap.

[lschumm]

I've done a mockup of some code to make this style of Scree plot:

import numpy as np
import matplotlib.pyplot as plt
import pandas as pd
from sklearn.decomposition import PCA

df = pd.read_csv("yellowbrick/examples/data/concrete/concrete.csv")

pca = PCA(n_components=9)
pca.fit(df)

plt.plot(np.arange(1, len(df.columns) + 1), np.cumsum(pca.explained_variance_ratio_))
plt.xlabel('number of components')
plt.ylabel('cumulative explained variance')
plt.xticks(np.arange(1, len(df.coulmns + 1), step=1))

[pdamodaran]

@lschumm - fyi - we have a visualizer to do this - it can be found here:

https://github.com/DistrictDataLabs/yellowbrick/blob/develop/yellowbrick/features/decomposition.py

[pswaldia]

As directed by @pdamodaran , there's already a scree plot visualizer present in yellowbrick for this purpose. I am closing this issue for now.

[BradKML]

Are there any Elbow-like tool for Determining optimal PCA count?

[bbengfort]

@BrandonKMLee I had been working on an explained variance plot for this - the WIP PR is here: #1037 -- it's been a while since I've taken a look at it, but if you want to help get it across the line or just use the code from that PR, I think that's what you're looking for.