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

Statistical Decision Making

Statistical decision baking is popular as data science method for risk.

Index for Decision Making

The basic index is expectation. Expectation is the sum of the product of productivity and effect (or loss). Expectation seems to be average with probability. It is the main index in Risk Evaluation .

Other indexes are maximin-standard, maximax-standard, Harwicz-standard, Laplace-standard and Savage-standard. Maximin-standard is calculated from the best number from the worst cases. It is called that "Maximin-standard is for pessimists. Maximax-standard is for optimists."

These indexes are difficult to use because we must prepare all data. In practical use, we do with limited data.

Decision Making with Bayesian Statistics

In the real world, the condition is changing one after another. But many methods of Data Science do not consider the changing condition.

Bayesian Statistics is good at the condition near the real. It is useful for below.

• To use the most of stored data.
• To change the output when new data is gotten.
• To use subjective knowledge effectively.

NEXT Psychology for Decision Making