Skip to content

A comprehensive predictive model to assist Saudi Aramco in reducing methane emissions as part of its sustainability efforts. The project integrates statistical analysis with advanced modeling techniques to forecast emission levels and suggest actionable steps for reduction.

Notifications You must be signed in to change notification settings

tyeckh/BC2406-Analytics-I-Visual-Predictive-Techniques

Repository files navigation

BC2406-Analytics-I-Visual-Predictive-Techniques

Methane, a potent greenhouse gas, has garnered global attention for its role in climate change. As part of international sustainability efforts, Saudi Aramco aims to reduce its environmental footprint. This project, conducted as part of the Nanyang Technological University Business School Coursework, serves as a Proof of Concept for Aramco.

Objectives

This project provides a data-driven solution to reduce methane emissions in the oil and gas industry, focusing on Saudi Aramco. Utilizing predictive modeling techniques, including Linear Regression, CART, and Random Forest, the project aims to forecast methane emissions and propose effective reduction strategies.

Three-Tiered Approach

Assessing Acceptability: Benchmarking predicted emission levels against industry standards.
Tackling Significant Variables: Identifying and strategizing actions to address key contributing factors.
Scenario Testing: Utilizing predictive models to optimize emission reduction strategies.

Key Findings

Technological upgrades, maintenance of older infrastructure, and efficient equipment deployment are critical to reducing methane emissions.

Proposed Solution

Our solution aims to position Aramco as a leader in sustainable energy practices, ensuring compliance with global standards. By lowering emissions, Aramco may reduce its cost of capital, enhancing shareholder value.

Conclusion

This project demonstrates the synergy between environmental sustainability and economic success. By actively addressing methane emissions, Aramco contributes to global climate efforts while improving its financial outlook.

Analytical and Predictive Techniques Utilised

  1. Logistic Regression was utilised in predicting class labels for missing catergorical data.
  2. Linear Regression was utlised in predictive modelling and the result was evaluated against the other models.
  3. Classification And Regression Tree (CART) was utlised in predictive modelling and the result was evaluated against the other models.
  4. Random Forest was utlised in predictive modelling and the result was evaluated against the other models.
    In addition, a web interface built on R Shiny was developed for easy reference and application to business reccomendations.

Repository Contents

Project Report: Sustainable_Methane_Management_Report.pdf
R Script (Modeling and Results): Methane_Management_Model.R
R Script (Web Application using R Shiny): Prototype_Modelling.R
Cleaned Dataset: Methane_Management_Dataset.csv

Acknowledgments

Group Members:
Ji Rong: Exploratory Data Analysis, Data Cleaning and Preparation
Kit Ho: Data Cleaning, Modelling and Evaluation
Yi Hong: Insights, Business Applications and Recommendations
Justinian: Business Recommendations and Conclusion

Course: BC2406 Analytics I

This project, including its report, script, and presentation, was developed solely as a Proof of Concept for academic purposes under the course BC2406 at Nanyang Technological University Business School. It was presented internally to the course lecturer and is not affiliated with or endorsed by Saudi Aramco or any commercial entities. All analyses, predictions, and recommendations are hypothetical and for educational purposes only.

About

A comprehensive predictive model to assist Saudi Aramco in reducing methane emissions as part of its sustainability efforts. The project integrates statistical analysis with advanced modeling techniques to forecast emission levels and suggest actionable steps for reduction.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages