Data analytics dashboard showing various charts and graphs

Projects, Case Studies & Articles

Welcome to my portfolio, a personal showcase of my journey and accomplishments in the dynamic world of data. This page is a reflection of my passion and dedication to transforming complex data sets into meaningful insights and innovative solutions.

Here, you'll find a collection of my projects, ranging from predictive analytics to advanced machine learning models, each illustrating my skills and expertise in data manipulation, visualisation, and interpretation.

Automating DLE Correction for Reservoir Engineering: A Python Based Approach to Enhance PVT Data Accuracy

This project automates the correction of Differential Liberation Experiment (DLE) data to field separator conditions using Python. The script imports and validates CCE, DLE, and Separator Test data from CSV files, then performs Bo and Rs corrections based on bubble-point pressure. Interactive plots are generated with Plotly for quality control, and results are exported to Excel. This automation reduces manual correction time from hours to minutes while improving accuracy for reservoir simulation and material balance calculations.

Tools used: Python (NumPy, Pandas, Plotly, Openpyxl). Click to view in GitHub.

Production Performance Dashboard with Marimo IDE

This project builds an interactive Production Performance Dashboard for oil and gas wells using Marimo IDE. The dashboard imports production, pressure, and well test data from Excel files, then computes key metrics including GOR, WOR, and their derivatives for diagnostic analysis. Features include multi-well selection, production allocation vs well test comparison, cumulative oil tracking, spatial well mapping, and configurable moving average plots. Interactive Plotly visualizations enable quality control and performance diagnostics, with data export functionality.

Tools used: Marimo IDE, Python (NumPy, Pandas, Plotly). Click to view in GitHub.

Fracture Width Prediction and Loss Prevention Material Sizing in Depleted Formations Using Artificial Intelligence

A Society of Petroleum Engineers (SPE) Published paper.

The core of the study involves using an Artificial Neural Network (ANN), a form of artificial intelligence, to develop a tool for predicting the width of fractures. The ANN is trained and validated with a limited dataset of 30 points, achieving a squared correlation of 79%. The findings from the ANN are then compared with existing 2D fracture models and benchmarked against experimental results found in literature. This approach represents a significant step in enhancing drilling efficiency and safety in mature oil and gas fields. Click to view publication.

Feasibility of Energy Substitution in Nigeria: Solar Panels and Inverter Batteries

“An Analysis of Economic Viability Under New Electricity Tariffs”

This paper evaluates the economic viability of substituting conventional energy sources with solar panels and inverter batteries in Nigeria, focusing on the impact of newly adjusted electricity tariffs in Nigeria for Ikeja, Lagos environs in 2024. The analysis employs Net Present Value (NPV) calculations over a 20-year period to compare solar, grid, and hybrid energy systems under varying financial and economic conditions. Click to read full article.

Financial Fraud Detection for a Bank's Loan Data using Machine Learning

In this project: I analysed the bank’s loan data and performed detailed Exploratory Data Analysis for missing values treatment & outlier removal. Performed Feature Engineering for the variables to be used in the Machine Learning Models. Predicted the tendency of their clients to default on loans with different Machine Learning Models and compared the Models’ performance.

Tool used: Python for visualisation and machine learning. Click to view in GitHub.

Analysing and Predicting Online Purchase Behaviour with Machine Learning

This project analyses an E-commerce customers data. Detailed Exploratory Data Analysis was done to pre-process the data, followed by Feature Selection to determine relevant variables, then ultimately 5 Machine Learning models were deployed in predicting the purchase tendency of the users. ML models deployed were: - Random Forest, Decision Tree, Logistic Regression, Gaussian NB & LinearSVC. Top 3 performing models for this dataset were the Logistic Regression, LinearSVC and Gaussian NB. Overfitting was observed in the other two models. Click to view in GitHub.

Data cleaning and analysis using SQL in Snowflake for the Loan Repayment Data of a Green Tech Company

This project shows the results following data cleaning and analysis using SQL in Snowflake for the Loan Repayment Data of a Green Tech Company in Germany.

I find querying data within Snowflake to be seamless, especially with the assisted functions to prompt overlooked analysis.

The SQL queries and presentation can be found in my GitHub repository. Click to view in GitHub.

Analysing Data Science Jobs and Salaries with Python

This study analyses job market data from 2020 to 2023, focusing on global salary trends, job title remunerations, and the impact of various factors on data-related roles. Utilising Python for data manipulation and visualisation, the study showed significant growth in the data sector, underscored the prominence and pay discrepancies of certain job categories, and highlighted geographical salary disparities. Click to view in GitHub.