Matt Mills
@statmills
Followers
413
Following
2K
Media
200
Statuses
2K
Data Scientist at Intuit/Mailchimp. I like to share random musings on R, Stats, and College Football
Atlanta, GA
Joined June 2010
I've uploaded some CFB Data for open source use https://t.co/YpY8nkZ8xH 10 years of Team recruit rankings Draft Picks Schedule and Results
0
3
22
10 of the 11 writers list FSU as making the playoff, none of them include Clemson at all, and yet at Fanduel right now Clemson is still the favorite to win the ACC at +185
sportsbook.fanduel.com
Bet on all college football games on FanDuel Sportsbook. Find college football odds for the biggest upcoming games.
College Football Playoff predictions: Who's most likely to make the field - via @ESPN App
0
0
0
@BudElliott3 FPI still has preseason projections built in, so even if teams play to their current ratings the changes from the preseason should get more drastic as the current season gets more weight.
0
0
0
Fun wrinkle for GT this year; only 3 conference opponents are playing better than expected. Technically FPI has us favored in every game until UGA lol @FTRSJoey
1
0
2
I'd guess you'd find the same results in basketball as well, the game the same it just got more fierce.
NEW with @KuperSimon The prevailing narrative around increased injuries and player workload in elite football is wrong. Players don’t play more football than in the past. What has changed is a sharp rise in intensity of play. Not more minutes, but each minute exerts more load.
0
0
0
Sharing for the morning crowd; My latest blog post covers how you can fit shape constrained models in python leveraging splines and JAX
My latest blog post is a walk-through of how Shape Constrained P-splines work and how you can use them to fit a curve of any arbitrary shape like monotonically increasing or decreasing #pydata #pystats #datascience #MachineLearning
0
0
0
This means you can enforce arbitrary shapes, even convex and concave, but still leverage all the benefits of a traditional GAM. Even better they are so straightforward you can fit them using general optimization packages like {jax} and {scipy}
1
0
0
They work by a simple yet effective reparameterization of a traditional GAM: 1. make all the coefficients positive 2. transform the coefficients with a running addition or subtraction of the previous coefficients to ensure they always get bigger or smaller
1
0
0
My latest blog post is a walk-through of how Shape Constrained P-splines work and how you can use them to fit a curve of any arbitrary shape like monotonically increasing or decreasing #pydata #pystats #datascience #MachineLearning
2
1
3
This lets you generate different levels of smoothing. There is an additional way to smooth the data using low rank smoothers. These not only smooth differences between neighboring values but also fit much faster by using fewer overall parameters.
0
0
0
The data was a chance for me to explore geo location models with {mgcv}. Similarly to penalizing the difference between neighboring coefficients in a GAM, you can penalize the difference between neighboring locations using a Markov Random Field
2
0
1
I'm excited to finally share some Atlanta Housing Data Charts! The first image shows the census tracts with the highest rate of first time buyers, the second shows the average sale price. Yes, they are basically inverses of each other but I enjoy the details of both
1
0
1
I hate how it's never the actual data analysis that trips me up when switching between R and Python, it's the silly base stuff like `int` and `len` that takes me multiple tries to switch over 🤬
0
0
0
Between him and Marcus Dupree there may be nothing cooler in CFB than an OU running back hitting a crease and seeing the crowd go insane for the long touchdown run
0
0
2
Especially the pysparklines package that can do multiple line graphs?? That is rrreeeaaalll nice
0
0
0
And some python packages already exist: https://t.co/mTxwFCkbxF & https://t.co/I0jPbZqG1v for much more robust implementations than mine.
pypi.org
Generate sparklines for numbers using Unicode characters only.
1
0
0
For the life of me I could not remember the proper name last night but some helpful redditors have reminded me these are called sparklines, from Tufte:
A couple of years ago I saw a really neat R function that printed a histogram as unicode text (▁▂▄█▆▃▁▁). I put together a python function that does the same so you can quickly see the distribution of your variables without creating tons of plots or for use in text blocks
1
0
0
You can view the code in this gist: https://t.co/ryoe9eRVyh Hope someone finds this helpful! #python #pystats #datascience #rstats
gist.github.com
A python function for printing a histogram as a unicode text string, e.g. '▁▂▄█▆▃▁▁' - display_hist.py
0
0
0