By scaling up the size of the sampling experiment, the credibility of the data can be greatly improved.
growth rate and growth rate
If a company's sales are:
100、101、102.5。
From the above data it can be concluded that sales first increased by 1% and then by 1.49% (1.5/101). Objectively speaking, this growth rate figure is very small. It can also be said that the latter growth rate is almost attached to the former data. In principle, we cannot use the word "vigorous" to describe the company's sales.
On the contrary, if we do not start from the initial data of growth rate, but instead use the growth rate of growth rate to express, then the situation will change fundamentally. If sales increase by 1% in year 1 and 1.49% in year 2 over year 1, then sales in year 2 are up 49% over year 1!
In short-term conditions, the trend can be used as a yardstick. (Cow milk production forecast example)
Sampling inspections have sample deformities (the United States conducted a weight survey in Germany after World War II, and the scales were placed in the square. Only those who were full could go out, and those who were not full would not go out. In fact, the standard standing on the scale has been filtered. shed most of the hungry people.)
The book also mentions survivorship bias
Be especially careful with the representation of digital graphics in higher dimensions (2D, 3D), as digital graphics often do not match dimensions.
The way you ask questions when surveying questions can lead to different feedback, jokes about smoking and praying. (If you want a sensible answer, then you must ask wisely - Goethe)