Feature Engineering - Introduction

Back to Index

Objective

Learn feature engineering

Reference

Essential Reading

• Why data prep is important
• Feature Engineering, how and why
• Chapter 5. Basic feature engineering of the book Real World Machine Learning

Extra Reading

• Fundamental Techniques of Feature Engineering for Machine Learning
• Feature Engineering exercises @ Kaggle

Checklist

After completing the exercises below, you should be comfortable with

• understand why feature engineering is very important in machine learning
• have a good overview of feature engineering tasks
• picking features that are relevant

Exercises

Difficulty Level

★☆☆ - Easy
★★☆ - Medium
★★★ - Challenging
★★★★ - Bonus

A - Outlier Detection

A-1 : House Sales Data (★★☆)

Read the house-sales-simplified.csv.
We are going to find outliers in sale price.

First, describe and visualize saleprice attribute. How can we know there are outliers?
Hint: You can look at standard deviation

We can eliminate top 10% and bottom 10% to find middle prices.

As next step, we segment house prices per bedrooms.

We also need to take zipcode into account when determining prices.

So our final assesment, we need to calculate prices per-bedroom, per-zipcode.
Come up with your assesment of outier detection.

More Exercices

• Pandas cleanup

---

Index