Lab 8: Model Building IV: Classification

Start Mar 19 –- Due Mar 24

Goals

In this lab, students will learn to:

• Build a classifier from a regression model

• Fit a generalized linear model (i.e., Logistic regression) to data

• Choose threshold to balance rates of false discoveries and false omissions

• Use undersampling to generate a balanced sample

• Estimate parameter uncertainties via K-fold cross validation.

• Train a neural network to classify data

• Improve performance of classifier by increasing size or number of hidden layers

• Test performance using a held-out test set

Instructions

• Work through the ex1.jl (Pluto notebook) and save updates in your personal repository.

• Work through the ex2.jl (Pluto notebook) and save updates in your personal repository.

• Commit and submit your work.

• Once the instrctor and/or TA has provided feedback, review the feedback

Resources

• GLM.jl

• Bayesian Logistic Regression

• Logistic regression with Flux.jl