Packt+ | Advance your knowledge in tech

You're reading from R Data Mining Implement data mining techniques through practical use cases and real-world datasets

Product type Paperback

Published in Nov 2017

Publisher Packt

ISBN-13 9781787124462

Length 442 pages

Edition 1st Edition

Languages

Tools

ggplot

Concepts

Data Mining

Author (1):

Andrea Cirillo

View More author details

Table of Contents (16) Chapters

Preface

1. Why to Choose R for Your Data Mining and Where to Start FREE CHAPTER

2. A First Primer on Data Mining Analysing Your Bank Account Data

3. The Data Mining Process - CRISP-DM Methodology

4. Keeping the House Clean – The Data Mining Architecture

5. How to Address a Data Mining Problem – Data Cleaning and Validation

6. Looking into Your Data Eyes – Exploratory Data Analysis

7. Our First Guess – a Linear Regression

8. A Gentle Introduction to Model Performance Evaluation

9. Don't Give up – Power up Your Regression Including Multiple Variables

10. A Different Outlook to Problems with Classification Models

11. The Final Clash – Random Forests and Ensemble Learning

12. Looking for the Culprit – Text Data Mining with R

13. Sharing Your Stories with Your Stakeholders through R Markdown

14. Epilogue

15. Dealing with Dates, Relative Paths and Functions

Measuring the performance in classification problems

Until now, we have just looked at regression settings, that is, problems where our main problem is to predict the level or value of a given response variable, starting from a set of explanatory variables. As you know, there are also classification problems, which are problems where you want to assign your observation to one of a given set of categories.

How do we measure the performance of these models? As always, you just have to resonate about the objective of your model to understand how to measure its performance. Our classification model aims to assign each observation to its category. How can you tell if it's doing well? You would probably count how many times it meets its objective, that is, how many correct classifications it performs.

This is actually one of the most common ways to a measure classification models' performance, even with some further development. Let's see it a bit closer.