Factual
How do you decide whether to use
kmeans
orkdemoids
?What is the significance of the boxplot layout? Why does it look that way?
Describe the underlying data produced in the outliers for the
iris
data, given the density plot.What are the extract rules for other items in the market dataset?
When, how, and why?
What is the risk of not vetting the outliers that are detected for the specific domain? Shouldn't the calculation always work?
Why do we need to exclude the
iris
category column from the outlier detection algorithm? Can it be used in some way when determining outliers?Can you come up with a scenario where the market basket data and rules we generated were not applicable to the store you are working with?
Challenges
I found it difficult to develop test data for outliers in two dimensions that both occurred in the same instance using random data. Can you develop a test that would always have several outliers in at least two dimensions that occur in the same instance?
There is a good dataset on the Internet regarding passenger data on the Titanic. Generate the rules regarding the possible survival of the passengers.