Question

We have a dataset where we're considering the rule "if Professor Kimmer then Disposition- Mean". There are 1000 instances in the dataset. 400 of them have Professor Kimmer (he gets around) and 600 of them have Disposition-Mean. 300 instances have both Professor-Kimmer and Disposition Mean. A. What is the confidence of the rule "if Professor Kimmer then Disposition Mean"? B. What is the support of the consequent "Disposition Mean"? C. What is the lift of the rule "if Professor Kimmer then Disposition Mean"? D. What is the likelihood that "Professor Kimmer"? (This is in the Naive Bayes section for this week.) E. What is the likelihood that "Disposition Mean"? what is the likelihood that "Professor = Kimmer AND Disposition Mean"? G. If "Professor - Kimmer" and "Disposition - Mean" were statistically independent, what would the likelihood of both occurring be? H. What is your answer to F divided by your answer to G? How does it compare to your answer to C? How does that support the interpretation of lift as how much more likely the items in the rule occur together than they would if they were statistically independent?

Answer 1

There are 1000 instances in the data set,

400 of them are Professor = Kimmer

and 600 of them are have deposition = mean

300 instances have both professor=Kimmer and deposition = mean

A)

the confidence of the rule " the professor = Kimmer then deposition is mean is given by "

$P(D=M|P=K)$

$P(D=M|P=K)=\frac{P(D=M\bigcap P=K)}{P(P=K)}$

300/1000 400/1000

$=\frac{3}{4}$

B)

the support of the consequent deposition= mean is given by,

PD-M)-600/1000 0.6