Stock Price Prediction Evaluation

This project is indented to present a small evaluation of different types of regression models for predicting stock prices for AAPL.

Setup

Steps:

• Download the repository
• Install packages: pip install -r requirements.txt
• Run: python comparison.py

Evaluation

Comparison

Models sorted by confidence

Method	Confidence	Mean Absolute Error	Mean Squared Error	Root Mean Squared Error
Orthogonal Matching Pursuit	99.9644%	12.944	251.349	3.598
Orthogonal Matching Pursuit CV	99.9644%	12.944	251.362	3.598
Linear	99.9641%	12.898	249.516	3.591
Bayesian Ridge	99.9641%	12.898	249.513	3.591
Ridge CV	99.9641%	12.897	249.494	3.591
Ridge	99.9641%	12.892	249.295	3.591
Elastic Net CV	99.9639%	12.876	248.647	3.588
Lasso	99.9634%	13.116	259.209	3.622
KNN	99.7029%	13.293	254.76	3.646
Elastic Net	99.4877%	12.879	218.512	3.589
Kernel Ridge	71.8265%	40.886	1778.924	6.394

Models sorted by RMSE

Method	Confidence	Mean Absolute Error	Mean Squared Error	Root Mean Squared Error
Elastic Net CV	99.9639%	12.876	248.647	3.588268
Elastic Net	99.4877%	12.879	218.512	3.588714
Ridge	99.9641%	12.892	249.295	3.590572
Ridge CV	99.9641%	12.897	249.494	3.591279
Bayesian Ridge	99.9641%	12.898	249.513	3.591345
Linear	99.9641%	12.898	249.516	3.591357
Orthogonal Matching Pursuit CV	99.9644%	12.944	251.362	3.597757
Orthogonal Matching Pursuit	99.9644%	12.944	251.349	3.597776
Lasso	99.9634%	13.116	259.209	3.621619
KNN	99.7029%	13.293	254.76	3.646022
Kernel Ridge	71.8265%	40.886	1778.924	6.394188

Moving Average for Adj Close

In the next paragraphs you can find predictions using selected methods.

Settings:

• Alpha: 0.1
• Number of iterations: 1000
• Alphas search: 0.0001, 0.0001, 0.001, 0.1
• Number of neighbors: 2