14.5 CONFIRMING OUR METRICS
We will call Rule ID [1] “Rule 1.” Next, let us confirm the following values for Rule 1, using what we have learned so far:
- Support,
- Confidence,
- Lift.
- Support.
How did we get the 36? Support requires the intersection of two events, which can be found by generating the contingency table of customer service calls vs churn, shown in Figure 14.5. Note that the cell for CSC = 5 and Churn = True contains Count = 36, represented 1.2% of the total number of records.
--s = s u p p o r t = P ( C S C = 5 a n d C h u r n = T r u e ) = transactions with both C S C = 5 and Churn = True total number of transactions = 36 3000 = 1.2 % - Confidence. Use the contingency table to confirm that this equals the conditional probability P(B ∣ A).
--c o n f i d e n c e = P ( C h u r n = T r u e | C S C = 5 ) = P ( C h u r n = T r u e a n d C S C = 5 ) P ( C S C = 5 ) = number of transactions containing both C S C = 5 and Churn = True number of transactions containing C S C = 5 Both...