At Kenya Agricultural & Livestock Research Organization (KALRO), I processed, cleaned, and standardized over 10,000 rows of agricultural consumption data, improving data accuracy by 15%. I also supported systematic data collection, entry, and validation across agricultural domain repositories.
I built a personal loan eligibility model using an Artificial Neural Network, applied SMOTE to address class imbalance, and evaluated its performance with confusion matrix and ROC-AUC measures. For a telecom customer churn project, I compared CART, Random Forest, and ANN models.
On a retail customer segmentation project, I used K-Means to create five behavioural profiles and translated the clustering results into targeted marketing recommendations. I also designed an interactive Tableau dashboard to track project implementation statuses, sites, and operational metrics across Kenya.
