TriPeaks Solver
During COVID-19 brain fog, I had just enough mental capacity to play a lot of Microsoft Solitaire games. I found that TriPeaks on master and grandmaster diff...
Posts by Tag
- R 28
- AI 18
- machine learning 16
- data visualization 15
- ggplot2 11
- regression 11
- python 7
- reinforcement learning 5
- neural network 5
- API 5
- college football 5
- nonparametric regression 4
- tensorflow 3
- OpenAI 3
- gym 3
- plotly 3
- design of experiments 2
- linear algebra 2
- tidymodels 2
- MCMC 1
- cryptology 1
- pytorch 1
- h2o 1
- GDELT 1
- geohash 1
- markdown 1
- GitHub Pages 1
- predictive modeling 1
R
Maze Generator
This generative art seems less artsy to me, but the idea of generating a maze was interesting. I found the mazegenerator GitHub repo that described the algor...
KNN Art
The generative art journey continues with images produced from machine learning models. In this post, I’ll use k-nearest neighbors to produce two types of ar...
Perlin Noise Flow Fields
In this post, I’ll generate art based on the idea of a material flowing through an area. Again this was inspired by the the aRtsy package, which has a canvas...
Space Colonization
Continuing the series of generative art posts, I liked the look of the canvas_petri() images from the aRtsy package. That led me to this article that describ...
The Recaman Sequence
The next stop on my generative art journey took me to the Recaman Sequence. As opposed to many other algorithms, there’s no randomness in this sequence - the...
The Collatz Conjecture
The inspiration for a series of posts on generative art was a post I read announcing the aRtsy package for producing generative art. A number of the example ...
Best Wordle Starting Word
I was scrolling through the RStudio blog page, and a post by Arthur Holtz caught my attention. He wrote some code to identify what the optimal starting word ...
Neural Network Regression
The following is a subsection of a tutorial I co-authored that I wanted to reproduce here to work out how to get the figures to render properly. Plus, it nev...
Ridge Regression Adjusted Statistics
The idea for this post started off as essentially a replication of this post but using R and Tidymodels instead of Python.
College Football Predictions
Continuing with the streak of posts about college football posts, here I finally get into developing some models to predict the winning team and margin of vi...
Plotly & Markdown
In a previous post, I created a scatterplot using ggplot2 that displayed college football team logos as the markers using geom_image() from the ggimage packa...
Simple Rating System
This, to me, is the coolest part. I had no idea this is how some rating systems work, and it’s pretty slick. It’s just one big system of equations that you s...
Oregon Football
I’m not a huge football fan. I can’t even remember the last time I watched an NFL game. However, I do sort of follow college football. And being from Oregon,...
Resolution 5 Fractional Factorial Designs
At work, I needed to create a resolution V (R5) fractional factorial design for use in a study of the relative effectiveness of various modernization programs.
Kaggle Housing Prices Competition
This notebook was written to document the steps and techniques used to achieve a top 2% ranking on Kaggle’s Housing Prices Competition.
Random Forests
As with support vector machines, and as the name implies, classification and regression trees (CART) can be used for either classification or regression task...
Support Vector Machines
Support Vector Regression (SVR) attempts to include as many data points as possible in the area between two lines. The following figure demonstrates this usi...
Generalized Additive Models
Recall that if there is a non-linear relationship between predictor and response, we can attempt to transform the predictor using a known function (log, reci...
Non-Parametric Analysis Of Variance
As we’ve seen in the last few posts, linear models can be successfully applied to many data sets. However, there may be times when even after transforming va...
Advanced Experimental Designs
So far, we have considered designs with just two levels per factor, which is fine for inherently binary factors. However, if our study requires us to include...
Logistic Regression
So far, we’ve seen regression techniques for continuous and categorical response variables. There is a different form of regression called logistic regressio...
Transforming Non-Linear Variables
If we have data that includes factors with more than two levels, we have the ability to evaluate non-linear relationships between predictor and response vari...
Model Selection
This post presents methods for finding a balance between under fitting and over fitting a model. Under fitting is when the model is a poor predictor of the r...
Multiple Linear Regression
In the previous section we considered just one predictor and one response. The linear model can be expanded to include multiple predictors by simply adding t...
Linear Regression Assumptions
There are four assumptions fundamental to linear regression:
Simple Linear Regression
The purpose of regression is to describe a relationship that explains one variable (the response, or the “y” variable) based on one or more other variables (...
COVID-19 Visualizations
I recently found the GitHub repository for the Johns Hopkins University Center for Systems Science and Engineering online dashboard, so I thought I’d do a li...
AI
TriPeaks Solver
During COVID-19 brain fog, I had just enough mental capacity to play a lot of Microsoft Solitaire games. I found that TriPeaks on master and grandmaster diff...
Neural Network Regression
The following is a subsection of a tutorial I co-authored that I wanted to reproduce here to work out how to get the figures to render properly. Plus, it nev...
Random Forests
As with support vector machines, and as the name implies, classification and regression trees (CART) can be used for either classification or regression task...
Support Vector Machines
Support Vector Regression (SVR) attempts to include as many data points as possible in the area between two lines. The following figure demonstrates this usi...
Generalized Additive Models
Recall that if there is a non-linear relationship between predictor and response, we can attempt to transform the predictor using a known function (log, reci...
Non-Parametric Analysis Of Variance
As we’ve seen in the last few posts, linear models can be successfully applied to many data sets. However, there may be times when even after transforming va...
Advanced Experimental Designs
So far, we have considered designs with just two levels per factor, which is fine for inherently binary factors. However, if our study requires us to include...
Logistic Regression
So far, we’ve seen regression techniques for continuous and categorical response variables. There is a different form of regression called logistic regressio...
Transforming Non-Linear Variables
If we have data that includes factors with more than two levels, we have the ability to evaluate non-linear relationships between predictor and response vari...
Model Selection
This post presents methods for finding a balance between under fitting and over fitting a model. Under fitting is when the model is a poor predictor of the r...
Multiple Linear Regression
In the previous section we considered just one predictor and one response. The linear model can be expanded to include multiple predictors by simply adding t...
Linear Regression Assumptions
There are four assumptions fundamental to linear regression:
Simple Linear Regression
The purpose of regression is to describe a relationship that explains one variable (the response, or the “y” variable) based on one or more other variables (...
PyTorch DQN Solves LunarLander-v2
A couple of weeks ago, I attempted to install the GPU version of TensorFlow and failed miserably. I should have set up a new virtual environment for it, but ...
Driving Up A Mountain Faster
The DeepMind researchers developed the Double Deep Q-Network after observing that the DQN target network is prone to overestimating Q-Values.
Driving Up A Mountain
A while back, I found OpenAI’s Gym environments and immediately wanted to try to solve one of their environments. I didn’t really know what I was doing at th...
Deep Q-Networks
In a previous post on Q-learning, I demonstrated how to train an agent to play tic-tac-toe. One of the challenges with Q-learning is that it doesn’t scale we...
Q-Learning
I’ve been reading some books on machine learning, and recently started going through Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow by Au...
machine learning
KNN Art
The generative art journey continues with images produced from machine learning models. In this post, I’ll use k-nearest neighbors to produce two types of ar...
Neural Network Regression
The following is a subsection of a tutorial I co-authored that I wanted to reproduce here to work out how to get the figures to render properly. Plus, it nev...
College Football Predictions
Continuing with the streak of posts about college football posts, here I finally get into developing some models to predict the winning team and margin of vi...
Kaggle Housing Prices Competition
This notebook was written to document the steps and techniques used to achieve a top 2% ranking on Kaggle’s Housing Prices Competition.
Random Forests
As with support vector machines, and as the name implies, classification and regression trees (CART) can be used for either classification or regression task...
Support Vector Machines
Support Vector Regression (SVR) attempts to include as many data points as possible in the area between two lines. The following figure demonstrates this usi...
Generalized Additive Models
Recall that if there is a non-linear relationship between predictor and response, we can attempt to transform the predictor using a known function (log, reci...
Non-Parametric Analysis Of Variance
As we’ve seen in the last few posts, linear models can be successfully applied to many data sets. However, there may be times when even after transforming va...
Advanced Experimental Designs
So far, we have considered designs with just two levels per factor, which is fine for inherently binary factors. However, if our study requires us to include...
Logistic Regression
So far, we’ve seen regression techniques for continuous and categorical response variables. There is a different form of regression called logistic regressio...
Transforming Non-Linear Variables
If we have data that includes factors with more than two levels, we have the ability to evaluate non-linear relationships between predictor and response vari...
Model Selection
This post presents methods for finding a balance between under fitting and over fitting a model. Under fitting is when the model is a poor predictor of the r...
Multiple Linear Regression
In the previous section we considered just one predictor and one response. The linear model can be expanded to include multiple predictors by simply adding t...
Linear Regression Assumptions
There are four assumptions fundamental to linear regression:
Simple Linear Regression
The purpose of regression is to describe a relationship that explains one variable (the response, or the “y” variable) based on one or more other variables (...
Q-Learning
I’ve been reading some books on machine learning, and recently started going through Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow by Au...
data visualization
TriPeaks Solver
During COVID-19 brain fog, I had just enough mental capacity to play a lot of Microsoft Solitaire games. I found that TriPeaks on master and grandmaster diff...
Maze Generator
This generative art seems less artsy to me, but the idea of generating a maze was interesting. I found the mazegenerator GitHub repo that described the algor...
KNN Art
The generative art journey continues with images produced from machine learning models. In this post, I’ll use k-nearest neighbors to produce two types of ar...
Perlin Noise Flow Fields
In this post, I’ll generate art based on the idea of a material flowing through an area. Again this was inspired by the the aRtsy package, which has a canvas...
Space Colonization
Continuing the series of generative art posts, I liked the look of the canvas_petri() images from the aRtsy package. That led me to this article that describ...
The Recaman Sequence
The next stop on my generative art journey took me to the Recaman Sequence. As opposed to many other algorithms, there’s no randomness in this sequence - the...
The Collatz Conjecture
The inspiration for a series of posts on generative art was a post I read announcing the aRtsy package for producing generative art. A number of the example ...
Best Wordle Starting Word
I was scrolling through the RStudio blog page, and a post by Arthur Holtz caught my attention. He wrote some code to identify what the optimal starting word ...
Ridge Regression Adjusted Statistics
The idea for this post started off as essentially a replication of this post but using R and Tidymodels instead of Python.
College Football Predictions
Continuing with the streak of posts about college football posts, here I finally get into developing some models to predict the winning team and margin of vi...
Plotly & Markdown
In a previous post, I created a scatterplot using ggplot2 that displayed college football team logos as the markers using geom_image() from the ggimage packa...
Simple Rating System
This, to me, is the coolest part. I had no idea this is how some rating systems work, and it’s pretty slick. It’s just one big system of equations that you s...
Oregon Football
I’m not a huge football fan. I can’t even remember the last time I watched an NFL game. However, I do sort of follow college football. And being from Oregon,...
GDELT Tutorial
GDELT claims to be the largest, most comprehensive, and highest resolution open database of human society ever created.
COVID-19 Visualizations
I recently found the GitHub repository for the Johns Hopkins University Center for Systems Science and Engineering online dashboard, so I thought I’d do a li...
ggplot2
TriPeaks Solver
During COVID-19 brain fog, I had just enough mental capacity to play a lot of Microsoft Solitaire games. I found that TriPeaks on master and grandmaster diff...
Maze Generator
This generative art seems less artsy to me, but the idea of generating a maze was interesting. I found the mazegenerator GitHub repo that described the algor...
KNN Art
The generative art journey continues with images produced from machine learning models. In this post, I’ll use k-nearest neighbors to produce two types of ar...
Perlin Noise Flow Fields
In this post, I’ll generate art based on the idea of a material flowing through an area. Again this was inspired by the the aRtsy package, which has a canvas...
Space Colonization
Continuing the series of generative art posts, I liked the look of the canvas_petri() images from the aRtsy package. That led me to this article that describ...
The Recaman Sequence
The next stop on my generative art journey took me to the Recaman Sequence. As opposed to many other algorithms, there’s no randomness in this sequence - the...
The Collatz Conjecture
The inspiration for a series of posts on generative art was a post I read announcing the aRtsy package for producing generative art. A number of the example ...
Best Wordle Starting Word
I was scrolling through the RStudio blog page, and a post by Arthur Holtz caught my attention. He wrote some code to identify what the optimal starting word ...
Ridge Regression Adjusted Statistics
The idea for this post started off as essentially a replication of this post but using R and Tidymodels instead of Python.
Plotly & Markdown
In a previous post, I created a scatterplot using ggplot2 that displayed college football team logos as the markers using geom_image() from the ggimage packa...
COVID-19 Visualizations
I recently found the GitHub repository for the Johns Hopkins University Center for Systems Science and Engineering online dashboard, so I thought I’d do a li...
regression
Neural Network Regression
The following is a subsection of a tutorial I co-authored that I wanted to reproduce here to work out how to get the figures to render properly. Plus, it nev...
Ridge Regression Adjusted Statistics
The idea for this post started off as essentially a replication of this post but using R and Tidymodels instead of Python.
Oregon Football
I’m not a huge football fan. I can’t even remember the last time I watched an NFL game. However, I do sort of follow college football. And being from Oregon,...
Kaggle Housing Prices Competition
This notebook was written to document the steps and techniques used to achieve a top 2% ranking on Kaggle’s Housing Prices Competition.
Advanced Experimental Designs
So far, we have considered designs with just two levels per factor, which is fine for inherently binary factors. However, if our study requires us to include...
Logistic Regression
So far, we’ve seen regression techniques for continuous and categorical response variables. There is a different form of regression called logistic regressio...
Transforming Non-Linear Variables
If we have data that includes factors with more than two levels, we have the ability to evaluate non-linear relationships between predictor and response vari...
Model Selection
This post presents methods for finding a balance between under fitting and over fitting a model. Under fitting is when the model is a poor predictor of the r...
Multiple Linear Regression
In the previous section we considered just one predictor and one response. The linear model can be expanded to include multiple predictors by simply adding t...
Linear Regression Assumptions
There are four assumptions fundamental to linear regression:
Simple Linear Regression
The purpose of regression is to describe a relationship that explains one variable (the response, or the “y” variable) based on one or more other variables (...
python
GDELT Tutorial
GDELT claims to be the largest, most comprehensive, and highest resolution open database of human society ever created.
PyTorch DQN Solves LunarLander-v2
A couple of weeks ago, I attempted to install the GPU version of TensorFlow and failed miserably. I should have set up a new virtual environment for it, but ...
Driving Up A Mountain Faster
The DeepMind researchers developed the Double Deep Q-Network after observing that the DQN target network is prone to overestimating Q-Values.
Driving Up A Mountain
A while back, I found OpenAI’s Gym environments and immediately wanted to try to solve one of their environments. I didn’t really know what I was doing at th...
Deep Q-Networks
In a previous post on Q-learning, I demonstrated how to train an agent to play tic-tac-toe. One of the challenges with Q-learning is that it doesn’t scale we...
Q-Learning
I’ve been reading some books on machine learning, and recently started going through Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow by Au...
Decrypting Substitution Cyphers
Stanford professor Persi Diaconis described an event when a psychologist from the California prison system brought encoded messages obtained from inmates to ...
reinforcement learning
PyTorch DQN Solves LunarLander-v2
A couple of weeks ago, I attempted to install the GPU version of TensorFlow and failed miserably. I should have set up a new virtual environment for it, but ...
Driving Up A Mountain Faster
The DeepMind researchers developed the Double Deep Q-Network after observing that the DQN target network is prone to overestimating Q-Values.
Driving Up A Mountain
A while back, I found OpenAI’s Gym environments and immediately wanted to try to solve one of their environments. I didn’t really know what I was doing at th...
Deep Q-Networks
In a previous post on Q-learning, I demonstrated how to train an agent to play tic-tac-toe. One of the challenges with Q-learning is that it doesn’t scale we...
Q-Learning
I’ve been reading some books on machine learning, and recently started going through Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow by Au...
neural network
Neural Network Regression
The following is a subsection of a tutorial I co-authored that I wanted to reproduce here to work out how to get the figures to render properly. Plus, it nev...
College Football Predictions
Continuing with the streak of posts about college football posts, here I finally get into developing some models to predict the winning team and margin of vi...
Driving Up A Mountain Faster
The DeepMind researchers developed the Double Deep Q-Network after observing that the DQN target network is prone to overestimating Q-Values.
Driving Up A Mountain
A while back, I found OpenAI’s Gym environments and immediately wanted to try to solve one of their environments. I didn’t really know what I was doing at th...
Deep Q-Networks
In a previous post on Q-learning, I demonstrated how to train an agent to play tic-tac-toe. One of the challenges with Q-learning is that it doesn’t scale we...
API
Ridge Regression Adjusted Statistics
The idea for this post started off as essentially a replication of this post but using R and Tidymodels instead of Python.
College Football Predictions
Continuing with the streak of posts about college football posts, here I finally get into developing some models to predict the winning team and margin of vi...
Plotly & Markdown
In a previous post, I created a scatterplot using ggplot2 that displayed college football team logos as the markers using geom_image() from the ggimage packa...
Simple Rating System
This, to me, is the coolest part. I had no idea this is how some rating systems work, and it’s pretty slick. It’s just one big system of equations that you s...
Oregon Football
I’m not a huge football fan. I can’t even remember the last time I watched an NFL game. However, I do sort of follow college football. And being from Oregon,...
college football
Ridge Regression Adjusted Statistics
The idea for this post started off as essentially a replication of this post but using R and Tidymodels instead of Python.
College Football Predictions
Continuing with the streak of posts about college football posts, here I finally get into developing some models to predict the winning team and margin of vi...
Plotly & Markdown
In a previous post, I created a scatterplot using ggplot2 that displayed college football team logos as the markers using geom_image() from the ggimage packa...
Simple Rating System
This, to me, is the coolest part. I had no idea this is how some rating systems work, and it’s pretty slick. It’s just one big system of equations that you s...
Oregon Football
I’m not a huge football fan. I can’t even remember the last time I watched an NFL game. However, I do sort of follow college football. And being from Oregon,...
nonparametric regression
Random Forests
As with support vector machines, and as the name implies, classification and regression trees (CART) can be used for either classification or regression task...
Support Vector Machines
Support Vector Regression (SVR) attempts to include as many data points as possible in the area between two lines. The following figure demonstrates this usi...
Generalized Additive Models
Recall that if there is a non-linear relationship between predictor and response, we can attempt to transform the predictor using a known function (log, reci...
Non-Parametric Analysis Of Variance
As we’ve seen in the last few posts, linear models can be successfully applied to many data sets. However, there may be times when even after transforming va...
tensorflow
Driving Up A Mountain Faster
The DeepMind researchers developed the Double Deep Q-Network after observing that the DQN target network is prone to overestimating Q-Values.
Driving Up A Mountain
A while back, I found OpenAI’s Gym environments and immediately wanted to try to solve one of their environments. I didn’t really know what I was doing at th...
Deep Q-Networks
In a previous post on Q-learning, I demonstrated how to train an agent to play tic-tac-toe. One of the challenges with Q-learning is that it doesn’t scale we...
OpenAI
PyTorch DQN Solves LunarLander-v2
A couple of weeks ago, I attempted to install the GPU version of TensorFlow and failed miserably. I should have set up a new virtual environment for it, but ...
Driving Up A Mountain Faster
The DeepMind researchers developed the Double Deep Q-Network after observing that the DQN target network is prone to overestimating Q-Values.
Driving Up A Mountain
A while back, I found OpenAI’s Gym environments and immediately wanted to try to solve one of their environments. I didn’t really know what I was doing at th...
gym
PyTorch DQN Solves LunarLander-v2
A couple of weeks ago, I attempted to install the GPU version of TensorFlow and failed miserably. I should have set up a new virtual environment for it, but ...
Driving Up A Mountain Faster
The DeepMind researchers developed the Double Deep Q-Network after observing that the DQN target network is prone to overestimating Q-Values.
Driving Up A Mountain
A while back, I found OpenAI’s Gym environments and immediately wanted to try to solve one of their environments. I didn’t really know what I was doing at th...
plotly
The Recaman Sequence
The next stop on my generative art journey took me to the Recaman Sequence. As opposed to many other algorithms, there’s no randomness in this sequence - the...
The Collatz Conjecture
The inspiration for a series of posts on generative art was a post I read announcing the aRtsy package for producing generative art. A number of the example ...
Plotly & Markdown
In a previous post, I created a scatterplot using ggplot2 that displayed college football team logos as the markers using geom_image() from the ggimage packa...
design of experiments
Resolution 5 Fractional Factorial Designs
At work, I needed to create a resolution V (R5) fractional factorial design for use in a study of the relative effectiveness of various modernization programs.
Advanced Experimental Designs
So far, we have considered designs with just two levels per factor, which is fine for inherently binary factors. However, if our study requires us to include...
linear algebra
Simple Rating System
This, to me, is the coolest part. I had no idea this is how some rating systems work, and it’s pretty slick. It’s just one big system of equations that you s...
Resolution 5 Fractional Factorial Designs
At work, I needed to create a resolution V (R5) fractional factorial design for use in a study of the relative effectiveness of various modernization programs.
tidymodels
Ridge Regression Adjusted Statistics
The idea for this post started off as essentially a replication of this post but using R and Tidymodels instead of Python.
College Football Predictions
Continuing with the streak of posts about college football posts, here I finally get into developing some models to predict the winning team and margin of vi...
MCMC
Decrypting Substitution Cyphers
Stanford professor Persi Diaconis described an event when a psychologist from the California prison system brought encoded messages obtained from inmates to ...
cryptology
Decrypting Substitution Cyphers
Stanford professor Persi Diaconis described an event when a psychologist from the California prison system brought encoded messages obtained from inmates to ...
pytorch
PyTorch DQN Solves LunarLander-v2
A couple of weeks ago, I attempted to install the GPU version of TensorFlow and failed miserably. I should have set up a new virtual environment for it, but ...
h2o
Kaggle Housing Prices Competition
This notebook was written to document the steps and techniques used to achieve a top 2% ranking on Kaggle’s Housing Prices Competition.
GDELT
GDELT Tutorial
GDELT claims to be the largest, most comprehensive, and highest resolution open database of human society ever created.
geohash
GDELT Tutorial
GDELT claims to be the largest, most comprehensive, and highest resolution open database of human society ever created.
markdown
Plotly & Markdown
In a previous post, I created a scatterplot using ggplot2 that displayed college football team logos as the markers using geom_image() from the ggimage packa...
GitHub Pages
Plotly & Markdown
In a previous post, I created a scatterplot using ggplot2 that displayed college football team logos as the markers using geom_image() from the ggimage packa...
predictive modeling
College Football Predictions
Continuing with the streak of posts about college football posts, here I finally get into developing some models to predict the winning team and margin of vi...