[PYTHON] DeepRunning ~ Level4.4.2 ~

Level4. Machine learning course (theory and practice)

What is a deep learning course that can be crushed in the field in 3 months

4-4. Principal component analysis

4-4-7. Hands-on

● Create a logistic regression model using breast cancer test data. ● Dimensional compression is performed on a two-dimensional space using the main component. (32 dimensions ⇒ 2D compression)

Unnecessary columns have been deleted properly.

Create objective variables and explanatory variables, and divide learning data and verification data.

As it is, it is 32 dimensions and the accuracy is 97.2%.

Analyze the main components, The axis of the first main component is 43% or more The axis of the second main component is about 20% The axis of the third main component is about 10% Therefore, it may be possible to maintain about 65% with the first and second main components.

⇒ Try to visualize it.

As you can see in the lecture, the boundaries are ambiguous in 2D.

4-4-8. Consideration

If you can pack data in everything and use many variables, the accuracy will increase. However, controlling the cost of calculation and maintaining accuracy will require experience and steady verification. I would like to try changing the explanatory variables to be adopted a little, and to accumulate failures and try. Principal component analysis is easy, but I felt it was a powerful analysis method.