The Need for Feature Engineering
Explore the necessity of feature engineering in data science, focusing on how transforming and creating features from raw data can improve the performance of machine learning models. This lesson explains key techniques like feature preprocessing and generation, using examples such as salary prediction and time series data to illustrate their impact on model quality.
We'll cover the following...
Feature engineering refers to generating and extracting features. Features play a significant role in solving any data science problems. Often our datasets are messy and contain improper unstructured data. To create an efficient model, we need to make data in a way that the model can understand and train it. In this section, you will learn the need for feature engineering and how we can benefit from it to build better models.
Consider an example of a salary dataset. We were given a few attributes of a person like age, sex, occupation, maximum education, working hours and we have to predict the salary range of a person. We will use a similar dataset in upcoming sections.
| Age | Sex | Occupation | Education | Working Hours | Salary Range |
|---|---|---|---|---|---|
| 20 | Male | Software Engineer | High-School | 20 | <40K |
| 35 | Male | Software Engineer | Post graduate | 50 | >80K |
| 30 | Female | Accountant |