---
library_name: sklearn
tags:
- regression
- scikit-learn
- UCS
- cement
license: mit
pipeline_tag: tabular-regression
---

# RandomForestRegressor for UCS Prediction

## Model Overview

This model is a `RandomForestRegressor` trained to predict the Unconfined Compressive Strength (UCS) of soil-cement mixtures based on the following features:

- **Curing Period (`curing_period`)**: Duration in days.
- **Compaction Rate (`compaction_rate`)**: Numerical value.
- **Cement Percentage (`cement_percent`)**: Percentage of cement in the mixture.

## Performance

The model achieved an R² score of **0.968** during cross-validation, indicating a high level of accuracy in predicting UCS values.

## Feature Ranges

- **Curing Period**: Min = 0.0 days, Max = 28.0 days, Mean = 11.06 days
- **Compaction Rate**: Min = 0.5, Max = 1.25, Mean = 0.989
- **Cement Percentage**: Min = 0.0%, Max = 10.0%, Mean = 5.77%

## Usage

To utilize this model, load the `model.joblib` file and input data within the specified feature ranges to obtain UCS predictions.

## Limitations

The model is calibrated for predictions within the specified feature ranges. Using it outside these ranges may result in less accurate predictions. It is specifically designed for soil-cement mixtures and may not be applicable to other materials.

## Author

Bogdan TEODORU
Gheorghe Asachi Technical University of Iasi (TUIASI)

## Contact

For inquiries or suggestions, please contact bteodoru@tuiasi.ro.