If you’ve worked in AI and machine learning, you probably know that getting good training data is one of the core problems with building good models. And in many cases, we need a lot of training data to build models that generalize well, meaning, models that perform well both during training, but also on new ... Read more ..read more

Dealing with the problem of overfitting is one of the core issues in machine learning and AI. Your model seems to work perfectly on the training set, but when you try to validate it on the test set … it’s terrible. This is a core problem in machine learning, especially when you start using advanced ML techniques like deep learning. And to solve the problem of overfitting, you often need regularization. Regularization is one of the primary tools that we use to build models that fit the data well, but also generalize well to unseen data. So in this blog post, I’m going to explain the essentials ..read more

If you want to build high-performing machine learning and AI systems, then simply training those systems is rarely enough. You often need to build multiple models, often with multiple different algorithms, and then compare the different models to each other to see which is best. And further, you often need to “tune” the settings of every different algorithm to get them to perform optimally (or close to it). That’s right. Most machine learning algorithms have “settings,” which we know in the industry as hyperparameters. Working with hyperparameters is a big part of machine learning. So in this ..read more

In machine learning, making sure that you have a model that performs well is, in some sense, the most important thing. This means that you need to be really good at evaluating different models. But, this can be a challenge, as you run into issues like overfitting, hyperparameter optimization, and model selection. One way to overcome these various model evaluation and selection challenges is with cross validation. Cross validation is a set of related techniques that we can use to evaluate and optimize our machine learning models. And, it’s a very important tool in the toolkit of a machine learn ..read more

Mmm. Overfitting. It’s the bane of most machine learning developers. You build a model that performs so well on the training data, and think “I’ve done such a good job!” while patting yourself on the back. Then, you test the model later, only to find that it performs terribly. Welcome to overfitting. Overfitting is probably one of the most important concepts to understand machine learning, because it’s at the core of building models that work in the real world. So in this blog post, I’m going to explain what overfitting is, what causes it, and a few high-levels ways to diagnose it. If you need ..read more

Welcome to our deep dive into one of the foundations of machine learning: Training, Validation, and Test Sets. In this blog post, I’ll explain the purpose of having these different machine learning datasets, explaining their roles, and discuss a few of the main strategies for data splitting. If you need something specific, just click on any of the following links. Table of Contents: Machine Learning Dataset Basics Why we need training, validation and test sets What are the Training, Validation, and Test Sets Basic Data Splitting Strategies Frequently Asked Questions Having said that, this bl ..read more

If you want to master machine learning and AI, you’ll need to learn and master a variety of minor concepts that underpin these systems. One such concept is the classification threshold. The classification threshold is critical for a large number of machine learning algorithms – from logistic regression to neural network classification – and it’s particularly important for binary classification tasks. Understanding classification threshold is important not only because model performance generally relies on the threshold setting, but more specifically, because the threshold determines performanc ..read more

In this tutorial, I’ll show you how to plot a precision-recall curve in Python using Seaborn. Specifically, I’ll show you how to use the new Seaborn Objects package to plot it. The article is divided into several different sections to make it easier to navigate. So if you need something specific, just click on any of the links below. Table of Contents: Quick Review of Precision-Recall Curves Setup Create Data Plot the Precision-Recall Curve with Seaborn Objects Why Use Seaborn for Precision-Recall Curves Frequently Asked Questions I’ll show you the actual example further down in the tutorial ..read more

Binary classification stands as a fundamental concept of machine learning, serving as the cornerstone for many predictive modeling tasks. At its core, binary classification involves categorizing data into two distinct groups based on specific criteria, a process akin to making a ‘yes or no’ decision. This simplicity conceals its broad usefulness, in tasks ranging from email spam detection to medical diagnosis. This blog post aims to demystify binary classification, clearly explaining what it is, key concepts, and a few real-world examples. If you need something specific, you can click on any o ..read more

In this blog post, I’m going to quickly explain positive and negative classes in machine learning classification. I’ll explain what the positive and negative classes are, how they relate to classification metric, some examples of positive and negative in real-world machine learning, and more. If you need something specific, just click on one of these links: Table of Contents: A Quick Review of Classification Defining Positive and Negative Real World Examples of Positive and Negative Classes How Positive and Negative Classes Relate to Classification Metrics That said, let’s dive in. We’ll sta ..read more