** Wed 25 January 2017

  || |Carbon dating, more specifically Carbon-14 dating (also sometimes called radiocarbon dating), is a technique used to determine the approximate age of old objects. It has some limitations; for instance, it can only be used on organic material, but it is such an important tool that the developer of the method, Willard Libby, was awarded the Nobel Prize for his work in 1960. His technique revolutionized the sciences of archaeology and paleontology.|| ||Radiocarbon dating was one of the most significant discoveries in 20th century science.The basic principle in radiocarbon dating is to measure the ratio of quanitity of the isotopes 14C to 12C that is present in a sample. We’ll see below how this is important:|

As I wrote in the previous post, I will continue in describing regression methods, which are suitable for double seasonal (or multi-seasonal) time series. In the previous post about Multiple Linear Regression, I showed how to use “simple” OLS regression method to model double seasonal time series of electricity consumption and use it for accurate forecasting. Interactions between two seasonal variables were successfully used to achieve this goal. The issue of forecasting time series from smart meters was discussed in my first post.

A couple weeks ago, I wrote about my initial efforts toward using machine learning to model the “master equations” that govern the Madden NFL player ratings system. This week, I’d like to put those models to use to compute the best player position switches for Madden 17.

Update, 3/14/2018: While I’ve still maintained the same basic workstation, I have done some upgrades. Before GPU prices skyrocketed, I added a 1080ti to workstation along with the 1060. However, this required an upgrade of my case to fit both. I ended up getting a Corsair 760T, which is an enormous increase in space over the Corsair mid-tower I had before.

I felt it was fitting to write the first “Project” article on this website, since it’s the most recent little project I’ve been working on.

Warning: This a rant post containing a bunch of unorganized thoughts.

The UCI hosts a dataset of wine measurements that is fantastic for demonstrating the importance of feature scaling in unsupervised learning . There are a bunch of real-valued measurements (of e.g. chemical composition) for three varieties of wine: Barolo, Grignolino and Barbera. I am using this dataset to introduce feature scaling my course on DataCamp.

I still haven’t heard from one of the 3 finalists, but I wanted to go ahead and post the first two, and I’ll update here with the final one later. These finalists win a data science book and a t-shirt, and I’ll choose from the three (I’m actually considering combining elements from two of them!) and announce the final t-shirt design when they are available for sale.

As of this writing, I’m in the middle of the transition from academia to industry data science. This website is meant partially to replace my academic website. It’s also a place for me to showcase some of my personal data science projects. I’ll probably be starting with some of the small projects I’ve done while building up some of my data science skills, and hopefully eventually be releasing some more polished projects.