Outliers are extreme maxima and minima far from the general level of numerical values. So it is also called outlier, and sometimes it is also called outlier.

The main reasons for the formation of outliers are: first, errors in sampling, such as recording errors, clerical errors of staff, calculation errors, etc. , may produce extreme value or extreme value. Secondly, it may be that the phenomenon under study itself is caused by various accidental abnormal factors. For example, in the population death sequence, due to an earthquake in a certain year, the number of deaths in that year increased sharply, forming an outlier; In the stock price sequence, stimulated by a certain policy or a certain rumor, there will be extreme rises and extreme falls, which will be realized as outliers.

No matter what causes the outliers, it will have a certain impact on the future analysis. Judging from the difficulties brought by the analysis, statistical analysts said that it is not desirable to have abnormal values in the series, which will directly affect the fitting accuracy of the model and even get some false information. Therefore, outliers are often regarded as "bad values" by analysts. However, from the information obtained, outliers provide very important information, which not only reminds us to carefully check whether there is an error in sampling, but also to confirm it carefully before analysis. Moreover, when it is confirmed that the abnormal value is caused by external sudden factors, it will provide important information such as system stability and sensitivity.