Soils are naturally formed materials and exhibit different properties that might alter depending on the field conditions. In order to determine the engineering behavior of soil as a construction form of foundation supporting material, it is required to examine the properties of soil using laboratory experiments. This process is challenging and time-consuming, and it requires the assistance of a competent technician. Knowledge of the compaction characteristics of a soil is perhaps the most important in determining the engineering properties of disturbed soils. This paper examines different statistical modeling strategies, specifically Multi Variable Regression Analysis and Artificial Intelligence Technique (Artificial Neural Networks) that are proposed for estimating the compaction characteristics namely Maximum Dry Density (MDD) and Optimum Moisture Content (OMC) based on basic soil properties like %Gravel(%G), %Sand(%S), %Fine Fraction(%FF), Modified Liquid Limit(LL)M and Modified Plasticity Index(PI)M of the Fine-Grained soils as the input parameters. 215 soils tested data, obtained from laboratory test results and literature review, were utilized in this study, the test data is divided into three parts: 70% is utilized for training the model, while the remaining 30% is split for validation (15%) and testing (15%) to evaluate the performance of models developed in this study. Comparison of observed and predicted values of output parameters values are performed using developed three models (Multi Linear Regression Model, Neural Network program Model and Neural Fitting tool Model) with same input parameters. The study reveals that the predicted values using neural network model are close to laboratory results and same is depicted from comparison of graphical plots of observed and predicted value of compaction characteristics for fine-grained soils.