Don’t wait to re-evaluate your analytic model
It can require a lot of time, money and effort to build a sound predictive analytic model. Between the data, analytics, development, testing, validation and production, it can easily take several months. Knowing this, it makes sense to use the model for as long as possible, right? As the old saying goes, “if it ain’t broke, don’t fix it.” Well, not necessarily.
Consumer behaviors are constantly shifting, reacting to market conditions. Changes at the macro level such as drifting consumer preferences or credit trends may be easier to detect, but lesser, more subtle changes are also occurring. In reality, it’s all happening, constantly evolving, and you can be sure it’s probably impacting the performance of your analytic models—and not in a good way. In other words, maybe the old saying should be revised to, “it might be broke, you just don’t know it yet.”
Frequently aligning your models with changing behaviors and conditions can help ensure that:
- Models remain theoretically and mathematically sound
- The data used in existing models is up-to-date and relevant
- Models are being applied correctly and performing as expected
- Results and interpretation are reliable with respect to the current economic environment
- Reporting results are accurate
But here’s the rub: keeping analytic models updated isn’t always easy. It can involve several internal teams going back and forth with each other and working with multiple sources of high-volume data, within multiple technology environments. By the time the update gets to market, it’s probably already a few months old. It’s why many organizations simply leave their models in place for a prescribed period of time—say 24 months—before addressing any possible shifts in performance.
Today, technology is changing all that, simplifying and speeding up everything. What if you had 24/7, real-time visibility into model performance, instead of having to follow a formalized redevelopment process and calendar? Better yet, what if you could query and access model performance data through an interactive online app via your laptop or tablet? What if those performance metrics were illustrated and easy to understand to help make the data more immediately actionable?
All these things are available now, through a dynamic, new analytic solution suite called Equifax IgniteTM. You get one-stop access to the diverse data and advanced technology you need to build and update analytic models—all within a single, secure cloud-based environment.
But, this is where Equifax Ignite is changing the game.. It also includes access to an online “app store” with several apps dedicated to model performance monitoring, redevelopment, validation and more. The apps offer instant clarity and visibility into model performance by delivering easy-to-understand data visualizations [think: graphs, charts, tables] that allow you to dynamically drill down deeper into the data with a single click versus filtering through 100’s of static spreadsheets and presentation slides.
Here’s the best part: you also create a direct feedback loop. The immediate, interactive nature of the apps creates a direct feedback loop that allows you to easily monitor, analyze and update your models, anytime. You simply click the app to submit a data request. In seconds, you get the data in an illustrated format so you can quickly spot issues, shifts or emerging trends. It’s fast, actionable insight that you can use to revise then re-fit an analytic model—all within the flexible, and secure Equifax Ignite environment—to keep it running at peak performance. You can always analyze the answer to the question: What lift can I get if we re-fit my models? And they are only deployed when you decide it’s time.
Don’t wait to evaluate your analytic models. Today it’s easier than ever before for you to keep your models current, relevant and optimized. A well-monitored model that reflects current market conditions, consumer behaviors and economic trends delivers a better result. If you’re interested in learning more about Equifax Ignite, please visit http://www.equifax.com/equifaxignite/.