an example of a feature may be color, with classes- red, blue, white, black, etc
I want to calculate some statistical measures out of a dataset, for e.g. skewness, kurtosis, etc. So, I want to encode the categorical features in a way that does not alter the statistical significance in any way.
You enter the design and data as collected, and the results are significant or not depending on reality (true relationship) and your sample size (estimated relationship).
There is always the risk that the way that you have asked your question and your experimental design influence the answer that you interpret from your data. If nothing else, sampling from an underlying population introduces uncertainty in the outcome and as the sample size declines the uncertainty grows exponentially.
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication