Last Updated: June 22, 2018

According to Dr. Keating, many enterprise companies are using outdated data science for forecasting methods. Dr. Keating, as a veteran in the space long before Big Data, was a common phrase, he can see why large companies with an abundance of accessible information are still missing the mark.

“Thirty years ago, the problem companies had was that very little data was available. Companies often didn’t keep track of sales, price changes or promotions – at least not two or three years in the past. Most of the models we used were time-series models … that have been around for 50, maybe 100 years, such as moving averages, exponential smoothing or ARIMA,” said Dr. Keating.

The reason time series models have been used for so long, he explained, is because they handle trends, seasonality, and cyclicality very well. However, those models are blind to changes in the economy.

“The federal reserve bank of St. Louis did a study four years ago and found that anytime there is a downturn in the economy, the forecast errors of the Fortune 500 quadruple! The reason is – they are using time-series models. They don’t have that ability to see ahead in terms of the economic changes,” Dr. Keating added.

Considering external data science is critical to predict what’s ahead

Ultimately, if the economy stays constant, then time-series models or ARIMA forecasting methods are appropriate, but that does not work for future-oriented outlooks. Dr. Keating explained, “Time-series models work very well when the economy is very stable. When the economy goes through a climate change, those models become totally unusable.”

He continued, “Now, what’s happening in the economy in the next 5 years? How many businesses are you aware of that were around in your childhood that aren’t around now? Things are changing very quickly, so as we get situations where it is easier to use outside data, time-series models will be quickly pushed in the background.”

Prevedere is a game changer.

Reflecting on Prevedere’s data science driven approach to forecasting, Dr. Keating explains how greater forecasting accuracy offers value across an organization “Notre Dame uses Prevedere in the college of business for students interested in going into analytics. That’s virtually every student, whether it is marketing, management, accounting, finance, every area uses analytics,” said Dr. Keating.

With Prevedere, decision-makers are able to leverage advanced forecasting methods such as pulling external data sources to see how they impact various functions, from finance to marketing. “The difficulty for most companies is figuring out how to use all this data all of the time,” said Dr. Keating. “But using an engine like Prevedere, they are going to find new uses for their data that they’ve kept in the past – uses that they’d never conceived when they began collecting the data.” 

Technology advancements will continue to evolve data science, forecasting and planning processes as solutions like Prevedere continue to innovate and empower companies to use data to their advantage. “It’s something that we haven’t seen before – there’s a familiar pool of software names that we use for statistics and mathematical correlations that fall into one category. Then there are public data sources like the Federal Reserve or StatUSA, available to everyone. Now if you can put the two together and add in proprietary software, provide a way to determine which pieces of data are important to forecast whatever it is you want to forecast – that’s where Prevedere is,” said Dr. Keating.

Business processes have changed and improved over time, but considering the importance of forecasting methods on performance, have been slow to evolve. Until now.

Common data science forecasting challenges like failing to consider the factors outside of your own four walls and not knowing what data to keep and dismiss, or how to interpret it, that’s where Prevedere changes the game. According to Dr. Keating, “Prevedere is a unicorn in terms of data science.”