Skip to content

Simulation is an incredibly useful tool in data science. We can use simulation to evaluate how algorithms perform against ground truth, and how algorithms compare to one another. In this project, I will be implementing and extending the nested spheres simulation study found in Elements of Statistical Learning page 339.

Notifications You must be signed in to change notification settings

MahtabEK/Nested-Spheres

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 

Repository files navigation

Nested-Spheres

Simulation is an incredibly useful tool in data science. We can use simulation to evaluate how algorithms perform against ground truth, and how algorithms compare to one another. In this project, I will be implementing and extending the nested spheres simulation study found in Elements of Statistical Learning page 339.

The Simulation Study

These are the steps I have followed in this project:

Steps:

  1. Generate a training data set of 2000 observations according to the description above. Label each of these training examples according to the rule above.

  2. Train a bagged estimator, a random forrest with max_features=1, a random forest with max_features=3, and an additional model of your choice (you can increase max features again, or you can explore a boosted estimator). Use 500 trees in your random forests and bagged estimator.

  3. Generate a testing data set of 10,000 observations according to the description above. Label each of these training examples according to the rule above.

  4. Use each model to predict on the testing data. Record the testing error rate (that is 1 - accuracy).

  5. Repeat these steps 50 times. Plot the error rates as a box plot by model to complete the assignment.

About

Simulation is an incredibly useful tool in data science. We can use simulation to evaluate how algorithms perform against ground truth, and how algorithms compare to one another. In this project, I will be implementing and extending the nested spheres simulation study found in Elements of Statistical Learning page 339.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published