cancel
Showing results for 
Search instead for 
Did you mean: 

Another interesting way to use a Payoff Matrix

AutoDR
DataRobot Employee
DataRobot Employee

Another interesting way to use a Payoff Matrix

Justify cost savings vs identify profit drivers


Robot 1:

Hello Team - A client and I were putting together numbers for a payoff matrix and had an alternative way to look at things. For them, the goal is to justify cost savings vs identify profit drivers from a use case.

 

Here's the existing docs: https://docs.datarobot.com/en/docs/modeling/analyze-models/evaluate/roc-curve-tab/profit-curve.html#...

 

Example:

  1. True Positive (benefit): This is the benefit from an order that is correctly predicted as canceled. The benefit is no/limited inventory costs. For example, if the item costs $100 to store typically, but due to cancellation - no additional cost (we put a 0 here). Benefit can come from additional revenue generated through proactive reach out
  2. True Negative (benefit): This is the benefit from an order that is correctly predicted as not canceled. The additional benefit / costs are 0 because a customer simply does not cancel this item and continues with shipment ($1000 profit on avg or -100 inventory cost per order)
  3. False Positive (cost): This is the cost associated with classifying an order as canceled when it did not cancel. What costs are incurred - Lost opportunity or business since the order is not fulfilled or delayed (-200)
  4. False Negative (cost): This is the cost associated with classifying an order as not canceled, when it will actually cancel. We have to incur a cost here for inventory management -$100

 

Just thought I'd share!

Labels (1)
0 Kudos
0 Replies