Monday, August 26, 2024
Media Contact: Tanner Holubar | Communications Specialist | 405-744-2065 | tanner.holubar@okstate.edu
Researchers from the School of Engineering, Architecture and Technology will begin a three-year project to use machine learning to measure and forecast China's agricultural production, inventory and imports.
A partnership between Oklahoma State University, Iowa State University and Cornell University, funded by the U.S. Department of Agriculture, has launched a project to understand the factors that influence China's imports and exports.
Guiping Hu, PhD, professor and dean of the School of Industrial Engineering and Management and Donald and Cathy Humphreys Chair, is helping lead the project as one of the principal investigators.
The project will focus on researching data and using machine learning to understand the factors that influence China's imports and exports. Another goal is to make policy recommendations and share findings with farmers and other stakeholders.
Machine learning is an aspect of AI research that leverages data and algorithms to improve a program's accuracy by mimicking how humans learn.
“The goal of the research is to utilize the data available through the import-export control system to understand some of the factors that affect import-export relations,” Hu said.
Hu said he pursues interdisciplinary research whenever possible, and this project brings together economists and engineers.
During his time at ISU, Hu used machine learning to study imports and exports in the mining and textile industries, which became the basis for a new project studying China's agricultural economy.
Traditional economic analysis models exist that can achieve these kinds of predictions; this project seeks to leverage innovative aspects of cutting-edge machine learning and data analytics tools to achieve the same results.
Machine learning techniques can improve the accuracy of models and predictions compared to traditional techniques, and also provide better insights when designing models.
The project uses high-quality data – carefully vetted, rigorously collected, structured databases – and some data that may not be included in databases, such as policy information or anecdotal information.
These types of data will be structured or incorporated to make them more informed, Foo said.
The dataset used was collected through collaboration between government agencies and non-profit organizations involved in international trade.
The team looks at historical data to explore how things have unfolded over time and uses machine learning to consider factors that could have led to different outcomes.
There are three levels of analysis in this project: descriptive, predictive, and prescriptive. Descriptive means using historical data to understand what happened and trying to understand relationships between data and trends.
Predictive analytics involves taking historical data, examining factors that have changed, understanding relationships between the data and trends, and making predictions about what will happen in the future.
Prescriptive analysis is about taking known facts about what has happened in the past and what could happen in the future, and looking at ways that can be changed to improve it. Hu notes that this involves asking a lot of “what if” questions that can be relevant to policymaking, policy advocacy, and scenario analysis.
The first phase of the project will focus on descriptive analysis of the data: the team will focus on collecting data and understanding some of the factors that may not be represented in the data.
“Even if you do all this comprehensive data analysis and getting all these types of data, there may still be data missing,” Hu said, “for example, some policies or certain USDA regulations may not be directly representative, so you may need to do what's called data engineering.”
The second phase focuses on descriptive and predictive analytics and examining historical trends related to the data, while the third phase focuses on predictive and prescriptive analytics., This includes analysing scenarios, developing policy recommendations and analysing possible “what-if” scenarios.
The main focus of the research is to make policy recommendations based on the findings.
“We try to tackle real-world problems and generate insights using data and some industrial engineering tools, so that stakeholders can benefit from our research and analysis,” Hu said.