I am glad you find this article useful. I have elaborated more for Point 4. "A cardinal variable is a numeric variable that its values are ordered. A non-cardinal variable, typically categorical, is a variable that its values cannot be ordered. For example, the value for the variable gender, "male" and "female", cannot ordered.

Suppose you need to convert a categorical variable, say gender, to a numeric variable to run a regression. Assume you assigned 1 to "male" and 2 to "female". You have mistakenly forced a categorical variable to be cardinal!

Below is a categorical "AP006" with values "h5", "android", "api", and "ios". I print out the count statistics for each value and the first 20 records (the first record is "android", the second is "h5", and so on.) There is no reason that the value "h5" is higher or lower than other values, i.e., no cardinal nature. If a new variable "newvar" is created as below, the value of the brand 'android' is absurdly two times higher than the value of 'h5'. This mistake should be avoided."