Concrete is one of the most widely used construction material in the world. For developed countries like Japan, the trend of concrete consumption is affected by significant changes over time. These changes can be attributed to the decrease in population, stability of the economy, declining need of new infrastructures and other hidden factors that might not be easily recognizable with conventional statistical modeling. Understanding these drivers of concrete material consumption is important in evaluating construction resource efficiency. As such, this paper aims to understand concrete material consumption trends in Japan by utilizing machine learning techniques. Machine learning has gained popularity mainly due to its self-learning characteristics that allows performance enhancement without being explicitly programmed. A backward approach of stepwise regression analysis was performed to quantify the contribution of socioeconomic factors to concrete consumption. After which, an agglomerative hierarchical clustering was made to identify similar patterns of concrete consumption behavior across the prefectures of Japan and group these prefectures together. Through the detection of patterns in the historical data, understanding the drivers of concrete consumption leads to the enhancement of the efficiency of resource consumption in the future.