Boise State University (BSU) doctorate students Evi Ofekeze and Ibrahim Olalekan Alabi have been focused on machine learning (ML) and data science under BSU Research Computing Services Director and West Hub Steering Committee Member Steve Cutchin for the past few years. Most recently the two students contributed to the NASA-funded SnowEx project with BSU Geoscientist H.P. Marshall using ML to predict hydrological outcomes of snowpack from remote sensing data.
Additionally, Ofekeze and Olalekan Alabi have been working with BSU Associate Professor of Marketing Anne Hamby and Assistant Professor of Information Technology Steve Pentland to analyze the varying degrees of success in response to fundraising videos on Kickstarter, a crowdfunding platform that often contains videos in which an entrepreneur describes their idea to potential investors.
“In the current work, we are examining how the vocal characteristics (in terms of pitch and loudness) associated with the entrepreneur's description predict the likelihood of obtaining funding,” explained Hamby. “We extracted the time series data from each video in terms of the pitch and loudness variation, and Evi and Ibrahim used BSplines to obtain the ‘shape’ of different videos on these dimensions - for instance, whether the pitch increases continuously over the video, increases then decreases, etc.”
They then used cluster analyses to group the main categories of patterns, from which three emerged for pitch and five emerged for loudness. Hamby said that some clusters are more persuasive than others in terms of their association with crowdfunding success.
“The work with the College of Business at BSU aligns with our goal of advancing the use of machine learning in various research domains,” said Ofekeze.
“We are also able to perform advanced data analytical techniques in social science research,” said Alabi. “This indicates the scalability of ML techniques.”
Ofekeze and Alabi are also currently working on their Software Carpentry trainer certifications.