ML Soil Suction Prediction Model

A machine learning model for predicting soil suction properties to aid geotechnical engineering decisions.

View Source

Role

ML Engineer & Researcher

The Problem

Traditional soil testing is time-consuming and expensive. Engineers need faster methods to estimate soil properties for construction projects.

The Solution

Developed ensemble ML models (Random Forest, XGBoost, Neural Networks) trained on 5,000+ soil samples. Built a web interface for engineers to input parameters and get predictions.

Outcomes

Achieved 94% prediction accuracy. Published research paper. Model used in 3 civil engineering projects.

Tech Stack

Python
Scikit-learn
XGBoost
TensorFlow
Flask
Pandas