Projects
Microalgae growth modeling
In this project, I developed a predictive model to evaluate the effect of various factors on the growth rate of microalgae in mass production. Using advanced machine learning algorithms, including LightGBM, XGBoost, and CatBoost, I built models to optimize the growth conditions. Before modeling, I conducted a thorough data preprocessing phase, which involved cleaning the data by handling duplicates and outliers, as well as encoding categorical variables and creating new features. I then used the trained models to determine the optimal values for each key feature, maximizing the efficiency of microalgae production.
​You can find the full project on GitHub: Here
Global Fisheries and Aquaculture Dashboard: Insights into Production, Species.
I created a Power BI dashboard that visualizes global fisheries and aquaculture data, including production by country, species, area, seas, and production type (capture or aquaculture). The interactive dashboard allows users to explore trends and make data-driven decisions to support sustainable fishing and aquaculture development.
Building a Predictive Model for Airplane Customer Booking Behavior
During my internship at British Airways, I developed a predictive model for airplane customer bookings using a dataset of 13 features, performing data cleaning, correlation analysis, and handling imbalanced data with techniques like SMOTE and Oversampling, undersampling. I trained a Random Forest model, tuned parameters, and evaluated its performance using ROC curves and AUC scores. While the model performed well, it showed some false negatives due to target imbalance.
You can find the full project on GitHub: Here
Transaction Data Analysis for Quantum
In this project, I conducted a comprehensive analysis of transaction data for Quantum, focusing on enhancing their sales strategy. Using R , I developed custom metrics and visualizations to provide actionable insights into customer purchasing behaviors and brand affinity. Applying the Pyramid Principle, I offered strategic recommendations for product placement and customer engagement. Additionally, I identified benchmark stores and suggested controlled experiments with trial layouts to evaluate their impact on sales and customer experience. This project strengthened my expertise in statistical techniques and data-driven decision-making, contributing to a more strategically aligned commercial approach for Quantum.
You can find the full project on GitHub: Here
Deep Learning Model for Microalgae Species Classification
In this project, we developed a deep learning model to classify microalgae species using images from the algaeBase database, working with a dataset of over 1,200 images per class. We employed the ResNet50 architecture, pre-trained on ImageNet, to utilize its robust feature extraction capabilities. To monitor training, we used Weights and Biases (Wandb) for real-time insights and adjustments. The model achieved high performance, demonstrating its effectiveness and reliability in species classification.
