It has been on the [return] the word is not quite understand why making predictions kind of problem, called regression problem, so he had checked the origin of this word.
The origin of the term
Regression analysis by the famous British anthropologist, statistician Galton (F.Galton, 1882 ~ 1911) founded. Years ago, when Galton was dedicated to research in the field of chemistry and genetics, he studied the relationship between father and son in the creation of the tall Englishman regression analysis.
In 1889, Galton and his students, one of the founders of modern statistics Pearson (Pearson) collected data on 1078 Height father and an adult son found after making a scatter plot based on survey data, sub high father does have a tendency to give birth to a son tall; the same sawed-off father tend to give birth to a son sawed.
After Galton study also found that his father's 1078 average of 68 inches tall, the average height of 1078 his son was 69 inches. A natural idea is: If the father's height is x inches, then the average height for his son's roughly x + 1 inch. But Galton in the study found that when the father is 72 inches tall, the average height of their son was only 71 inches, and did not reach 73 inches. He also found that the average height of only 64 inches tall and father whose son was 67 inches, even 65 inches higher than 2 inches expected. Thus he concludes: Despite the high father often give birth to the son of the high, low birth father loves most short son, but "sons," the tall man had wanted to trends in the average height of all close, that is to have the average height offspring "return" tendencies.
Galton's explanation is: to have a binding nature, so that the distribution of human height in a relatively stable period of time without producing polarization, which is called the "return" effect. Through this example, Galton introduced the "return" (reversion, then slowly evolved into regression) is used.
Now, regression analysis has been widely used in all aspects of scientific research 䣌 become one of the most important methods to explore relationships between variables, and to identify the specific form of relationship between the variables of a causal relationship.
In a nutshell
All in all, the return of the word, the relationship between a set of data is described, that data closer to the average. (All high school textbook knowledge of the forgotten ....)
Reference material
- Know almost https://www.zhihu.com/question/30123729