Should you make your decisions on a WhIM? Data-Driven Decision making using a What-If Machine for Evaluation of Hypothetical Scenarios

TL;DR

This paper introduces a What-If Machine that uses data resampling to evaluate hypothetical scenarios, supporting real-time, data-driven decision making across various tabular datasets by automatically identifying impactful areas.

Contribution

The paper presents a novel real-time What-If Machine that automatically highlights high-impact areas on target metrics, surpassing previous methods by enabling broader applicability and faster analysis.

Findings

Supports real-time hypothetical scenario analysis

Automatically identifies high-impact areas on target metrics

Applicable to any tabular data without specific use-case constraints

Abstract

What-if analysis can be used as a process in data-driven decision making to inspect the behavior of a complex system under some given hypothesis. We propose a What-If Machine that creates hypothetical realities by resampling the data distribution and comparing it to the an alternate baseline to measure the impact on a target metric. Our What-If Machine enables both a method to confirm/reject manually developed intuitions of practitioners as well as give high-impact insights on a target metric automatically. This can support data-informed decision making by using historical data to infer future possibilities. Our method is not bound by a specific use-case and can be used on any tabular data. Compared to previous work, our work enables real-time analysis and gives insights into areas with high impact on the target metric automatically, moving beyond human intuitions to provide data-driven insights.