This repository contains Jupyter Notebooks with R scripts inspired by the book Applied Regression Modelling by Iain Pardoe.
The book is mostly focused on the mathematical foundations of these models as opposed to implementation in R or Python, but in this repository I have created Jupyter Notebooks with R code that gives examples of how to build multiple linear regression models, interpret the result, check model assumptions, perform transformations and interpret influential points. These essential data science tasks are worked through along side mathematical formulas in the notebooks to show the underlying maths of each step.
- Model Comparisons, Assumptions & Predictions in Cars City Miles Efficiency
- There is value in visually examining the data, don't just compare the model params
- Compare Models, Check Assumptions & F-Test with Mortality & Air Pollution
- RSS, F-Test & Anova in predicting box office success
- Illustrating unimportant predictors with shipping labour hours dataset
- Transformations into Quadratic and Square Root Models
- Prediction Intervals and transformations for home tax dataset
- Transformations for GDP~Internet Model
- Analysis of interactions in multivariate analysis
- Removal of interaction terms
- Confounding levels in qualitative factors
- Diagnostic Plots, Leverage, Cooks Distance & Outliers
I am running a Jupyter Notebook server on my AWS EC2 instance accessible at https://stats.fieldmap.me/. If you would like access to the server please contact hi@johnmalcolmdesign.com . Datasets are hosted on S3.