This study presents a correlation of laboratory test results like particle size distribution, atterberg limits, specific gravity, and water content with compaction properties, i.e., maximum dry density (MDD) and optimum moisture content (OMC). A total of 450 tests were performed on 50 soil samples collected from the Konkan region of Maharashtra. Further, the test results of 55 soil samples are obtained from the literature. Levenberg Marquardt (LM) along with Scaled Conjugate Gradient (SCG) are the two algorithms of neural networks used for the prediction of interrelationships. Maximum dry density and optimum moisture content data of 105 soil samples were trained with respective index properties of the soil. Optimization on three different sets of inputs along with two algorithms was carried out. A reliable prediction model with four inputs and a coefficient of determination (R2 = 0.973) was obtained. The developed prediction model from this study shows a strong correlation between particle size distribution and compaction properties. Practical applications of this neural network model along with its limitations are also discussed in the paper.