Summary

In this chapter, we have learned about various techniques for dimensionality reduction. Let's summarize what we have learned in this chapter.

At the beginning of the chapter, we were introduced to the challenges inherent with some of the modern-day datasets in terms of scalability. To further learn about these challenges, we downloaded the Internet Advertisement dataset and did an activity where we witnessed the scalability challenges posed by a large dataset. In the activity, we artificially created a large dataset and fit a logistic regression model to it.

In the subsequent sections, we were introduced to five different methods of dimensionality reduction.

Backward feature elimination worked on the principle of eliminating features one by one until no major degradation of accuracy measures occurred. This method is computationally intensive, but we got better results than the benchmark model.

Forward feature selection goes in the opposite direction as backward...