notes on normality test gathered from machine learning mastery website

in machine-learning •  4 years ago 

Here is the source:
https://machinelearningmastery.com/a-gentle-introduction-to-normality-tests-in-python/

1- The parametric models assume that the data has Gaussian distribution. If they are not, we should use non-parametric models.
2- However, there are some middle points that we can transform the data into a Gaussian form. They are not just noise, and they are not completely Gaussian.
3- There are graphical tests and statistical tests. In statistical tests, we use methods to see how likely it is to transform it into a Gaussian model.
4- We can plot a histogram of data to see if it is like a bell or not, therefore if it is Guassian.
5- But it is better to use a statistical test. It will give us a "statistics" number and a "p-value". The statistics is not handy for use masturbators, but the p-value is.
6- So, if the p-value is greater than another number named "alpha", we can say that the data is not Guassian. Otherwise, it is. Alpha is likely to be 0.05 or percent.
7- The p-value of more than 5 percent means that "our data is likely to be normal.
8- So, here is a method to perform this test. It is good for lazy dumbasses like me to have fun: scipy.stats.normaltest .
(Yeah, I ain't gonna write my own test. Who do you think I am?!)

Authors get paid when people like you upvote their post.
If you enjoyed what you read here, create your account today and start earning FREE STEEM!