Top Page | Upper Page | Contents | About This Site | JAPANESE

Statistics

Statistics has long history. It is very useful. But we need time to get used to the ideas

• Normal Distribution and Others : To watch uneveness
• Average and Standard Deviation : taking care of meanings
• Hypothesis Testing : Methods to judge the difference
• Estimation : To predict the range of prediction.
• Information Theory : How we estimate "rare"?
• Bayesian Statistics: Use experience

True Average

We cannot get the value of true average exactly. But statics depends on the idea of true average. So statistics is mysterious study. The ideas of Hypothesis Testing and Estimation are made by this value.

By the way, I found that systematic errors are not clear in the real process control . So discussion of true average is not important in this analysis.

And there are cases that "we should have many true averages" or "Average changes dynamically". The analysis methods using one average are not used directly for these cases.

Truth and Sampling

Statistical values (average etc.) are calculated values. So quality of data is important. The keyword is " sampling ."

For example, if I studied the opinions of my friends then I report as the opinion of all people in Japan, it is a lie.

Many people point out similar cases of this example as lies.

It might be lies. But the first person gives the stage of discussions. So the report should be evaluated at this point. This idea is one of the heuristics.

NEXT Strength and Weakness of Big Data