Check whether the sample obeys the normal distribution, processing the bias distribution _ the number of places

In data analysis, if a certain data obeys a normal distribution, we can make a lot of meaningful analysis, such as T-Test, by using the nature of normal distribution.

How to check whether the sample obeys normal distribution.
You can use the Q-q graph for testing, q-q graph is a scatter plot, dot (x, y) represents a number of points in the data x, and y represents the same number of decimal places (i.e., FX (x) =fy (y)) as x, if the two-distributed QQ figure is on a line, then each [Xi,xi+1],[yi, The yi+1] interval contains the same proportion of data in the entire dataset, which means that if you indent x or y you can let them.

Therefore, if the unknown data and the standard normal distribution do q-q graph, if a bit in a straight line then the distribution of unknown data obeys the normal distribution.

What if it's not a normal distribution.
Data on the right side can be all the data logarithm, square root, etc., its principle is because the derivative of such a transformation is gradually reduced, that is, its growth gradually slowed down, so you can move large data to the left, so that the data near the normal distribution.
If the left side of the words can be converted to the reverse of the case of the right side.