Machine Learning and Classification

Name: Machine Learning and Classification
Author: Briana Brownell

Learn the basics of classification modeling and how to manage data collection and preprocessing

Preview the first lesson free — get full access to all 6 lessons.

Course: On-Demand

In this chapter, we will talk about the basics of classification modeling, including how it fits into the larger picture of machine learning and artificial intelligence. So, what is machine learning? I love this definition from Robert Schapire at Princeton because it gives us an intuitive way to think about ways that computers can learn. The first thing to notice in this definition is that all machine learning depends on different kinds of computer algorithms that we can use in order to solve different problems. The second thing is the dependency on data to make machine learning work. Now, because we live in such a data-rich world right now, it's easy to forget that this was not always the case. In history, data was actually quite scarce, and so most of the classification modeling and most of the AI modeling was actually not based on data observations; it was based on other forms of data that we could use to build some of these more complex systems, things like instruction and direct experience. But now we do have access to all kinds of data in all kinds of different domains. And so most of the machine learning modeling that you will do depends on learning from observational data. There are three main types of machine learning: supervised learning, unsupervised learning, and reinforcement learning. And all of these broad styles of machine learning solve different problems. For supervised learning, it's all about doing some kind of prediction or classification, and for that you need examples that are, what are called, labeled. And that means that you have a series of input variables and you're trying to use those to predict some kind of outcome. So if you think of the example of customer churn, you may have a bunch of data that looks at how satisfied that customer was with a product, how they were using that product, and then the outcome variable was whether or not they canceled their service. For supervised learning problems, it's essential to have that labeled variable that you can predict. By comparison, unsupervised learning does not have a labeled output variable, and so what that means is the kinds of problems that you're solving with unsupervised learning are a little bit different. You're looking at finding kinds of emerging patterns within the data that let you explore and use the data in a more sophisticated way; things like understanding what variables are related to each other; understanding things like different groups, different clusters within the data that you can use to understand the data that you have. And finally, reinforcement learning is a little bit different from both supervised and unsupervised machine learning. In reinforcement learning, there is an active agent that is learning a method to make decisions. So this style of machine learning is really common in games, so things like teaching and artificial intelligence how to play checkers or chess or any other game like that. Supervised machine learning is by far the most common that you will encounter in industry. And in fact, classification modeling is one important type of supervised machine learning. That said, unsupervised machine learning will also become extremely important even while you're doing supervised learning because a lot of the methodologies that you can use in unsupervised learning can be extremely important in doing things like feature selection and to understand the nuances within the data that you are using within a supervised machine learning problem, so using them both together is quite common. And so we will touch on some concepts of unsupervised learning in this course as well. Now you've got the basics of the three major types of machine learning and how supervised machine learning and classification model fits within them.

Intermediate Provider : Briana Brownell 6 Lessons · 26m in Arabic, German, English, Spanish, French, Portuguese, Chinese

Course Description

These lessons will introduce classification modeling and its problem space within supervised machine learning. This will include learning and recognizing the difference between classification and prediction, and how classification modeling fits in the larger field of machine learning. You will learn how to approach classification problems, and you will be introduced to the typical classification model pipeline.

Once you understand the classification model pipeline, you will learn how to begin the initial steps of the data collection. Finally, you will learn practical tips on how to manage data collection and information on preprocessing for a classification model.

What You'll Learn

Distinguish classification from prediction and how classification fits within the larger field of machine learning
Recognize basic classification concepts within supervised machine learning
Apply the typical classification model pipeline to approach classification problems
Begin the initial steps of data collection for a classification model
Manage data collection and preprocess data for a classification model

Key Takeaways

Classification modeling sits within the problem space of supervised machine learning.
Classification and prediction are distinct, and the course teaches how to tell them apart.
A typical classification model follows a pipeline, and understanding it helps you approach classification problems.
Data collection and preprocessing are important initial steps in building a classification model.
The course covers practical tips for managing data collection for classification.

Frequently Asked Questions

What does this course cover?

It introduces classification modeling and its problem space within supervised machine learning, including the difference between classification and prediction, how classification fits in the larger field of machine learning, the typical classification model pipeline, and the initial steps of data collection and preprocessing.

What lessons are included?

The lessons are Supervised Machine Learning, Basic Classification Concepts, Classification vs. Prediction, The Classification Model Pipeline, Data Collection, and Data Preprocessing.

What skills will I gain?

You will build skills in Data Classification, Machine Learning, Machine Learning Model Training, Machine Learning Model Monitoring and Evaluation, Statistical Classification, and Supervised Learning.

Does the course explain the difference between classification and prediction?

Yes. The course teaches how to learn and recognize the difference between classification and prediction and how classification modeling fits in the larger field of machine learning.

Does the course address data collection and preprocessing?

Yes. After covering the classification model pipeline, it teaches the initial steps of data collection, plus practical tips on managing data collection and information on preprocessing for a classification model.