03. Text: Lesson Topics

PCA Lesson Topics

There is a lot to cover with Principal Component Analysis (or PCA). However, you will gain a solid understanding of PCA by the end of this lesson, by applying this technique in a couple of scenarios using scikit-learn, and practicing interpreting the results.

We will also cover conceptually how the algorithm works, and I will provide links to explore what is happening mathematically in case you want to dive in deeper! Here is an outline of what you can expect in this lesson.

1. Dimensionality Reduction through Feature Selection and Feature Extraction

With large datasets we often suffer with what is known as the "curse of dimensionality," and need to reduce the number of features to effectively develop a model. Feature Selection and Feature Extraction are two general approaches for reducing dimensionality.

2. Feature Extraction using PCA

Principal Component Analysis is a common method for extracting new "latent features" from our dataset, based on existing features.

3. Fitting PCA

In this part of the lesson, you will use PCA in scikit-learn to reduce the dimensionality of images of handwritten digits.

4. Interpreting Results

Once you are able to use PCA on a dataset, it is essential that you know how to interpret the results you get back. There are two main parts to interpreting your results - the principal components themselves and the variability of the original data captured by those components. You will get familiar with both.

5. Mini-project

Finally, you will put your skills to work on a new dataset.

6. Quick Recap

We will do a quick recap, and you will be ready to use PCA for your own applications, as well as the project!