[PYTHON] Supervised learning (regression) 1 Basics

Aidemy　2020/10/28

Introduction

Hello, it is Yope! I am a liberal arts student, but I was interested in the possibilities of AI, so I went to the AI-specialized school "Aidemy" to study. I would like to share the knowledge gained here with you, and I am summarizing it on Qiita. I am very happy that many people have read the previous summary article. Thank you! This is the first post of supervised learning (regression). Nice to meet you.

• This article is a summary of what you learned in "Aidemy" "in your own words". It may contain mistakes and misunderstandings. Please note.

What to learn this time ・ About supervised learning (regression) ・ Linear regression method

Supervised learning (regression)

(Review) Machine learning method

・ There are three methods for machine learning. "Supervised learning," "unsupervised learning," and "reinforcement learning." ・ Of these, supervised learning can be divided into two categories: “classification” and “regression”. The regression we learn this time predicts continuous values such as __stock prices and market prices. __

Linear regression

What is linear regression?

-Linear regression is __ that predicts the movement of the data after that from the formula (graph = linear) of the data that is already known. The graph at this time is always a straight line (linear function).

Coefficient of determination

-The coefficient of determination __ indicates how well the data predicted by linear regression and the actual data match. -In scikit-learn used for supervised learning, the coefficient of determination takes a value between __0 and __. It can be said that the larger the value, the higher the degree of agreement. -When outputting the coefficient of determination, use __print (model.score (test_X, test_y)) __.

Linear simple regression

-Linear simple regression is a linear regression __ that obtains one data (y) to be predicted from one data (x). -That is, it means that __a and b are inferred, assuming that the data can be represented by "y = ax + b". -For guessing a and b, one is the __least squares method __. This is to set a and b such that __ the sum of the squares of the difference between the actual y and the estimated y (that is, ax + b) is minimized. -The reason why the difference is squared by this method is that it is not necessary to consider the positive or negative of the data.

-To use linear (simple) regression, use __ "model = LinearRegression ()" __.

-Execute linear simple regression

Linear multiple regression

-__ Linear multiple regression __ is a linear regression in which one data (y) is to be predicted and multiple data (x1, x2 ...) __ are used for the prediction. .. -That is, it means guessing a and b, assuming that the data can be represented by "y = a1x1 + a2x2 ... + b". -Even in linear multiple regression, the prediction of a and b is performed by the least squares method. -Also, you can use LinearRegression (). It automatically determines whether it is simple regression or multiple regression.

-Perform linear multiple regression

Summary

-Linear regression, which is one of supervised learning (regression), predicts the movement of data after that from an already known data formula (graph = linear). -Linear regression includes "linear simple regression" which predicts one and "linear multiple regression" which has multiple predictions.

This time is over. Thank you for reading until the end.