Non-Parametric Method
Categories: Metrics
Basically, the Get Out of Jail Free card for statistical analysis.
A non-parametric method is any statistical process or calculation that doesn’t require us to meet certain requirements in order to use it.
More specifically, processes like hypothesis tests and confidence intervals and correlation coefficients all require we have met a bunch of conditions, like knowing that the distribution of data is Normal in shape, or that the sample taken is random, or that the sample size is large enough.
A non-parametric method doesn’t need any of that jazz to be true for us to use it. Also, non-parametric methods are great for data that are “yes/no” or “married/single” in nature, and for data that need to be ranked, like customer rating data.