Date of Award

Summer 2026

Access Type

Thesis - Open Access

Degree Name

Master of Science in Civil Engineering

Department

Civil Engineering

Committee Chair

Siddharth Parida

Committee Chair Email

paridas@erau.edu

First Committee Member

Ashok Gurjar

First Committee Member Email

gurjara@erau.edu

Second Committee Member

Scott Kirts

Second Committee Member Email

kirtss@erau.edu

Third Committee Member

Stephen C. Medeiros

Third Committee Member Email

medeiros@erau.edu

Fourth Committee Member

Ryan Shamet

Fourth Committee Member Email

ryan.shamet@unf.edu

College Dean

James W. Gregory

Abstract

This thesis develops a machine-learning framework for estimating the compression index and the recompression index of Florida soils from routinely measured index properties, and reports two studies that build it. Consolidation settlement design requires both indices, and both are obtained from the incremental-loading oedometer test, which occupies a specimen for one to two weeks; the index tests that accompany it are complete within hours. Empirical correlations have filled that interval since the 1950s, but their coefficients are calibrated on specific soil populations and transfer poorly between regions. The first study analyzes 376 consolidation tests compiled for the Florida Department of Transportation. Ten index variables are ranked by Pearson and distance correlation, and polynomial and random-forest regressions trained on all ten are compared against models trained on the four highest-ranked; the reduced random forest reproduces the full one, at a coefficient of determination of 0.94 for fine-grained compression-index prediction. The second study expands the database to 497 records and generalizes the method: polynomial, ridge, support-vector, and random-forest regressions are compared under declared hyperparameters, with the random forest reaching 0.85 for the compression index and 0.70 for the recompression index; feature influence is ranked by Pearson correlation, distance correlation, and Gini importance; graduated pruning removes five low-importance variables without measured loss; and a conventional classification partition is compared against an unsupervised one, the 55-record coarse-grained stratum failing on both targets where both k-means clusters remain viable. Across the two studies and a 32% database expansion the same five variables govern prediction: initial void ratio, moisture content, dry unit weight, liquid limit, and plasticity index. The framework produces screening estimates that are partition-aware and five-input, and that remain subordinate to project consolidation testing.

Download

Share

COinS