August 13, 2021

In The Straits’ Times Ask NUS Economists series, Assistant Professor Denis Tkachenko (NUS Economics) explains how machine learning technology, commonly employed by technology giants such as Netflix and Google, makes the inter-disciplinary leap to solve real-world economics problems in policymaking, analysis, and evaluation.

The term ‘machine learning’ has gained traction in recent years, owing its popularity to IBM Marketing to attract clients and talented employees. While the term ‘machine learning’ has increased its popularity, the practice of machine learning itself is hard to define. Dr. Tkachenko broadly defines machine learning as the development of algorithms applied to data sets with the objective of prediction or clustering tasks.

Machine learning algorithms use statistical methods to build models based on sample data to make predictions or decisions. These models are constantly refined by ‘grading’ their predictive accuracy and introducing larger data sets. In a way, the algorithms are ‘learning’ how to make better predictions and classifications with each refinement.

Machine learning offers policymakers and economists unprecedented insight into forecasting economic trends and evaluating government policy. Dr. Tkachenko raises two examples of how machine learning techniques have already been applied to public policy. Firstly, machine learning algorithms performed substantially better than the models used by central banks in forecasting global economic conditions after the first year of the COVID-19 pandemic. Secondly, at home, Singapore’s Ministry of Trade and Industry employed machine learning algorithms, discovering a skills mismatch phenomenon in the domestic labor market. The Ministry’s findings informed policymaking decisions to diversify sources of information on jobs posted online for job seekers.

Although the use of machine learning has been closely associated with the technological domain, Dr. Tkachenko points out that these emerging technologies have far-reaching applications in other domains such as public policy, and that the rapid adoption of these technologies signal their growing importance in a digital economy based on data-driven decision making.

Read the article here!