GitHub Repositories

Machine Learning from “Scratch”

Building machine learning methods from basic principles is a great way to understand the mathematical and algorithmic intuitions behind the methods. Yes scikit-learn and such are quicker and easier (and generally faster), but it’s always good to have an understanding of what is going on “under the hood”.

This repository will be my examples of a variety of machine learning methods and algorithms from “scratch” (i.e., using minimal or base/common libraries). Check back, as I’ll be expanding the list as I have time and as I’m exploring more on my own.

Currently available:

k-means clustering

Coming soon:

Isomap
Random Forest

Data Science in Archaeology

A “day in the life” repository of using data science as an archaeologist… samples from some of my work projects and experiments (mainly in R). Much of this work revolves around simple ETL and EDA, summarization and descriptive statistics, manipulating and summarizing spatial data, and a few more “advanced” experiments with probability distribution modeling.

The code in this repository is very rough for now. I’ll clean them up into formal scripts as I have time.

Currently available:

Summarization utility functions for artifact counts, find depths, and temporally diagnostic dates using tidyverse
Experiments with creating mixture distributions (using mixdist) for temporally diagnostic artifact dates comprised of the independent uniform distributions for each artifact type. Working on bootstrap estimates.
Example automation code for generating standardized summaries for different artifact categories and spatial units for use in reports, piped to markdown tables using tidyverse. Incorporate spatial data pulled form ESRI shapefiles with terra and sf.