James D. McCaffrey

Bottom line: I compared my from-scratch C# implementation of gradient boosting regression on training data with categorical predictor variables with the scikit-learn GradientBoostingRegression library module and got essentially the same results.

Data smoothing is a classical statistics technique that removes noise from a dataset so that overall patterns and trends are visible.

I contributed some technical content and opinions to an article titled “Can a Single Entrepreneur Create a Successful AI-Based Startup Company?”

Zoltar is my NFL football prediction computer program.

Suppose you have a tiny dataset of nine items like this: Sex Hand Age Height City Score (y) ------------------------------------------- M right 24 tall Anaheim 0.60 F left 49 short Anaheim 0.70 M right 33 medium B…

Bottom line: Encoding categorical predictor variables for use in tree-based regression systems is a surprisingly complex topic.

As usual with complex problems, explaining what the problem is, is more difficult than explaining the solution.

I wrote an article titled “Neural Network Bistratum Regression Using C#” in the October 2026 edition of Microsoft Visual Studio Magazine.

Zoltar is my NFL football prediction computer program.

The form of a linear regression prediction equation is y’ = (w0 * x0) + (w1 * x1) + . . + (wn * xn) + b where y’ is the predicted value, the xi are predictor values, the wi are constants called model weights, and b is a constant called the bias.

One afternoon, I set out to come up with an algorithm for decision tree regression that is as simple as possible.

There were two editions of the Outer Limits TV series.

As usual with complex problems, explaining what the problem is, is more difficult than explaining the solution.

Zoltar is my NFL football prediction computer program.

RBF networks for regression had a brief period of popularity in the 1990s.

One evening, for no reason I can really think of, I decided to implement a quadratic regression system, using relaxed Moore-Penrose pseudo-inverse via QR-Householder inverse, from scratch, using JavaScript.

One Saturday morning, after I walked my two dogs (Riley and Kevin) and a visiting dog (Llama), I figured I would take a look at the XBGoost (“extreme gradient boosting”) regression code library.

Arguably the most basic form of machine learning regression is linear regression.

Zoltar is my NFL football prediction computer program.

I was working with tree-based regression systems.

The goal of a machine learning regression problem is to predict a single numeric value.

Technically, a deep neural network is one that has two or more hidden layers of processing nodes.

This animation always reminds me of a person who is very special to me — beyond special. 3W 5W OZ SZ The post For Someone Special appeared first on James D.

I contributed some technical content and opinions to an article titled “The AI Productivity Paradox” on the September 2026 edition of the Pure AI web site.

Zoltar is my NFL football prediction computer program.

I fed my decision tree regression system, implemented using from-scratch Python with NumPy, to several AI systems and asked the AI to analyze it for correctness.

The goal of a machine learning regression problem is to predict a single numeric value, for example, predicting the bank account balance of a person based on his age, annual income, and so on.

I decided to implement Gradient Boost regression using Blind Trees learners.

I wrote an article titled “Quadratic Regression with QR-Householder OLS Solve Training Using C#” in the September 2026 edition of Microsoft Visual Studio Magazine.

My standard from-scratch implementation of kernel ridge regression (KRR) trains using the Cholesky inverse of a RBF kernel matrix.

Zoltar is my NFL football prediction computer program.

There are three main ways to train a quadratic regression model: 1.) using stochastic gradient descent, 2.) using left pseudo-inverse (normal equations) via Cholesky inverse, 3.) using relaxed Moore-Penrose pseudo-inverse via one of many possible inverses.

I recently implemented support vector regression (SVR) from scratch using the C# language, trained using stochastic sub-gradient descent (SSGD).

Bottom line: I decided to implement AdaBoost.R2 regression using Extra (“Extremely Randomized”) Trees learners.

Zoltar is my NFL football prediction system.

One Sunday evening, I was sitting in my living room.

I like to write code every day.

The goal of a machine learning regression problem is to predict a single numeric value.

AdaBoost.R2 (“adaptive boosting regression, version 2”) is a technique to predict a single numeric value.

Zoltar is my computer program that predicts the results of NFL football games.

In general, when training a machine learning regression model using stochastic gradient descent, I do not use early-exit from the training loop.

Machine learning quadratic regression is a technique to predict a single numeric value.

I am a big fan of early science fiction movies of the 1950s.

I contributed some technical content and opinions to an article titled “AI Hallucinations: Why AI Confidently Lies and the Business Opportunities It Creates” on the Pure AI web site.

Zoltar is my NFL football prediction computer program.

A simple decision tree for regression will almost always overfit the training data where the model accuracy on the training data is near 100% but accuracy on new, previously unseen data is very poor.

The goal of a machine learning regression problem is to predict a single numeric value.