Predictive Analytics Essential Training: Estimating and Ensuring ROI

Course details

Nothing is more important to the future of predictive analytics teams than proving their projects have long-term value. Measuring the return on investment (ROI) often can help turn analytics into a visible profit center for your organization. Estimating ROI early—before a project even begins—can also help fast-track approval. Here Keith McCormick shows how to address ROI both before and after the predictive model is built. Learn how to create your estimate before the project starts by estimating the overall size of the problem, assigning value to possible outcomes, and judging the impact of model performance. Keith then shows a different method for calculating ROI after the model is built, during the evaluation and deployment phases, and provides tips for the ongoing monitoring of the project. He also takes a retrospective look assessed one year after model deployment. These two strategies will give you the data you need to get buy-in for your projects and provide ongoing metrics on their performance.

Concepts

0. Introduction

01 - Introduction

1. Effective Problem Definition

02 - Estimating ROI
03 - Business considerations when measuring ROI
04 - ROI starts with problem definition
05 - Why estimating ROI is perceived to be challenging

2. Estimating ROI before the Project Starts

06 - Introducing the confusion matrix
07 - The possible outcomes when a micro decision is made
08 - Estimating the overall size of the problem
09 - Assigning value to the four possible outcomes
10 - The impact of model performance on ROI
11 - Two other example scenarios

3. Calculating ROI after the Model Is Built

12 - Determining the cut-off on propensity scores for interventions
13 - A B testing and partial rollout during the evaluation phase
14 - Understanding the monitoring phase
15 - Revisiting ROI 12 months after deployment

Conclusion

16 - Next steps