Knowledge Discovery for Economic Forecasting: A Machine Learning Approach to Predict Labour Market Performance

The economists and policymakers are keenly interested in understanding the complex indicators of labour market performance to increase economic stability. However, the traditional econometric models that identify these indicators often struggle to capture the non-linear and complex relationships between different socioeconomic factors. This study addresses this gap by applying a machine learning approach to predict the unemployment and employment rates across OECD countries. The dataset containing different socioeconomic factors was downloaded from the OECD Data Explorer. After merging and cleaning the data, the decision tree and random forest models were trained. The results showed that the random forest model achieved an excellent accuracy, with an R-square of 0.98, for predicting the employment rate, while the unemployment rate proved to be more challenging to predict. The findings of this study also revealed that the level of education was observed as the main predictor of unemployment and employment, giving a robust empirical validation for human capital theory. This study determines that machine learning is the right approach for interpreting complex labor market dynamics. It provides a data-driven intervention mechanism for policymakers, highlighting that long-term investment in human capital is the fundamental strategy. So, policymakers can improve their policies using the key data-driven insights.