To improve the mechanical rationality of road pavement, it is essential to evaluate the compaction properties of the fine aggregate that constitutes the skeleton of the pavement as a soil element. Hence this study clarified the relationship between physical properties such as grain size distribution and compaction properties. To describe the compaction phenomenon of granular materials, the authors employed nine grain size feature values such as uniformity coefficient and fine fraction content, four water-related feature values including water content, and degree of saturation, wet density and compaction energy as feature values. Using these feature values based on the physical properties of fine aggregate, we attempted to perform feature analysis with Random Forest. The results show that the degree of compaction is highly correlated with the compaction energy and the degree of saturation. Based on the feature values, PCA is performed to clarify the feature values that affect the compaction properties.