In regression analysis, relation between Y and X is expressed by regression equation. Y is quantity data.

I think regression analysis is good as the start to study Multi-Regression Analysis because the idea in this method is used in many other methods and there are many applications ( Time Series Analysis etc.)

- Single Regression Analysis : Scatter Plot and Correlation , Faults When We Find Correlation , Application of Suspected Correlation , Correlation Analysis for Multi-Variable
- Multi-Regression Analysis : , Selection of Variance , Linearity , Partial Correlation , Pass Analysis

"I want to express the relationship between X and Y." and "I want to predict Y from X." are that many people expect. Some people may made some functions in their own way.

When such people (including me) learn Regression Analysis, they are glad to meet the method. "It's that I am looking for!"

But regression analysis has limits. There are many cases that it is not useful. In bad cases, we may lead a wrong result.

Regeression Analysis is good at the analysis of linearity between X and Y in static phenomena. Except Passing Analysis (an analysis of Y and time) , Self Correlation Analysis (an analysis of Y on time axis) and well- Stratified Analysis , Regression Analysis is not good at the the analysis of dynamic phenomena.

For the needs, "How I express the relationship between X and Y." and "I want to predict Y from X.", Meta Knowledge and the approach of Data Mining are also useful.

The memo, "
Prediction by Statistical Model

NEXT Correlation

Tweet