Future Forecast

While professional analysts have always had access to resources and data to predict future events in their industry, M.B.A. student Yogesh Singh Katoch is helping to bring that capability to, well, everyone.

Katoch, who is expected to graduate in May from the School of Management, is working with Shaya Sheikh, Ph.D., associate professor of supply chain and business analytics, as a research assistant on improving accuracy and speed of various kinds of forecasting models—from stock market trends and housing prices to election results and Academy Award winners.

The research merges traditional forecasting models with machine learning models to create more accurate predictions for stock prices and high-profile events using real-time data-driven analysis. Using a chatbot feature, anyone—whether someone with a vested interest in the information or just plain old curiosity—will be able to input a text command and find out which stocks are most likely to perform well or which football team is most likely to win the next NFL Super Bowl.

“This model combines traditional forecasting with modern deep learning methods and machine learning techniques to analyze historical data and current market conditions to make predictions,” Katoch says. “Chatbot users can put in the stock market values, and the model will show you results. You don’t have to go through the code or know how it’s working. You just put the values in, and it gives you the results.”

Katoch notes that the forecasting model can be adapted to analyze a wide variety of datasets, allowing it to predict the results of many types of events. “This flexibility means that the forecasting models and chatbot interface can continue to provide valuable, interpretable insights even when the context shifts from stock prices to another high-stakes event.”

In another machine learning forecasting project, Katoch is assisting with Sheikh’s research on a housing price prediction system, helping with data processing, model refinement, and evaluation of the forecasting results. The ability to predict housing prices in real time with high accuracy, Katoch explains, has many applications, including for home buyers and sellers, real estate agents, financial institutions, urban planners, insurance companies, and others.

Katoch is also building an interactive tool to help businesses improve their construction bidding strategies.

“For example,” Katoch says, “there are different construction bids in the market. How do you know what the perfect price for that bid is so you can get the construction contract and have no financial loss on the project? I built the model for the bid optimization calculator as my class project.”

With this research and technical experience boosting his prospects in the job market, Katoch looks forward to his career after graduation, hopefully as a financial analyst.

Katoch’s mentor has been pleased with his growth as a researcher as well as his future potential.

“Yogesh’s critical thinking and programming skills in progressing our housing price prediction research and other projects have been exceptional,” Sheikh says. “His ability to translate ideas into practical applications has consistently impressed me.”