🛢️ Machine Learning Based Well Log Analysis Dashboard

A Python-based interactive dashboard for well log analysis, petrophysical evaluation, and machine learning-driven lithology prediction using the FORCE 2020 Well Log Dataset.

The application enables geoscientists and petroleum engineers to upload LAS files, visualize well logs, compute petrophysical properties, predict missing density logs (RHOB), classify lithology using Random Forest models, and export processed results through an interactive Streamlit interface.

---

Project Overview

This project demonstrates an end-to-end digital well log interpretation workflow by integrating:

• Well log preprocessing
• Petrophysical analysis
• Feature engineering
• Machine Learning
• Interactive visualization
• Streamlit deployment

The workflow closely resembles industry practices used in reservoir characterization and formation evaluation.

---

Features

Well Log Processing

• Upload LAS files
• Automatic parsing using LASIO
• Missing value handling
• Interactive data preview

---

Petrophysical Analysis

The application computes:

• Volume of Shale (VSH)
• Density Porosity (PHID)
• Total Porosity (PHIT)
• Water Saturation (SW)
• Hydrocarbon Saturation (SH)
• Reservoir Quality Index (RQI)

using standard petrophysical equations.

---

Feature Engineering

Generated features include:

• LOG_RDEP
• LOG_DTC
• GR_RHOB
• GR_NPHI
• RHOB_NPHI
• POR_DIFF
• RQI

These engineered attributes improve machine learning model performance.

---

Machine Learning

RHOB Prediction

Random Forest Regression model for predicting missing density logs.

Lithology Classification

Random Forest Classifier trained on FORCE 2020 dataset for lithology prediction.

---

Interactive Dashboard

Visualize:

• Gamma Ray (GR)
• Density (RHOB)
• Neutron Porosity (NPHI)
• Deep Resistivity (RDEP)

using interactive Plotly charts.

---

Export

Download processed well log data as CSV.

---

Dataset

FORCE 2020 Machine Learning Competition Dataset

The dataset contains multiple North Sea wells with:

• GR
• RHOB
• NPHI
• DTC
• PEF
• RDEP
• Lithology labels

Raw dataset is not included in this repository due to licensing and file size limitations.

---

Repository Structure

Well-Log-Analysis/
│
├── app.py
├── README.md
├── requirements.txt
├── .gitignore
│
├── notebooks/
│
├── src/
│
├── figures/
│
├── reports/
│
└── data/

---

Technologies Used

• Python
• Streamlit
• Pandas
• NumPy
• Scikit-learn
• Plotly
• LASIO
• Joblib

---

Installation

Clone the repository

git clone https://github.com/UtkarshRode/Well-Log-Analysis.git

Move into the project

cd Well-Log-Analysis

Create virtual environment

python -m venv .venv

Activate virtual environment

Windows

.venv\Scripts\activate

Install dependencies

pip install -r requirements.txt

Run the application

streamlit run app.py

---

Workflow

LAS File

      │

      ▼

Data Loading

      │

      ▼

Cleaning

      │

      ▼

Petrophysical Analysis

      │

      ▼

Feature Engineering

      │

      ▼

RHOB Prediction

      │

      ▼

Lithology Prediction

      │

      ▼

Interactive Dashboard

      │

      ▼

CSV Export

---

Results

The application provides:

• Interactive well log visualization
• Petrophysical property estimation
• Missing RHOB prediction
• Lithology classification
• Downloadable processed results

---

Future Improvements

• XGBoost and LightGBM models
• Facies classification
• Multi-well comparison
• Cross-plots
• Pay zone detection
• Formation tops detection
• Cloud deployment
• Model explainability using SHAP

---

Author

Utkarsh Rode

Dual Degree (B.S. + M.S.) in Geology

Indian Institute of Technology Kharagpur

GitHub: https://github.com/UtkarshRode

---

License

This project is intended for educational and research purposes.

The FORCE 2020 dataset belongs to its respective owners and is not redistributed in this repository.