Live
Binary Classification · Insurance
Insurance Claim Prediction
Predict fraudulent or high-cost insurance claims before processing. SHAP explains every prediction to non-technical underwriters. Stops revenue loss at the source.
Random Forest · SMOTE · SHAP · 7,014 records
PythonScikit-learnSHAPSMOTEStreamlit
Live
Binary Classification · HR Analytics
Yakub Staff Promotion Prediction
Remove bias from HR promotion reviews at Yakub Trading Group. Flag high-potential employees with data, not gut feel. F1=0.411, AUC=0.891.
Random Forest · F1=0.411 · AUC=0.891
PythonPandasScikit-learnStreamlit
Live
Binary Classification · Banking
Bank Customer Churn Prediction
Identify customers about to leave before they actually do. Turning a reactive retention problem into a proactive strategy. F1=0.609, AUC=0.868.
Gradient Boosting · F1=0.609 · AUC=0.868
PythonGradient BoostingScikit-learnStreamlit
Live
NLP · Text Classification (3MTT Capstone 2)
TruthLens — Fake News Detector
Automatically detect misinformation in news articles. My first NLP project and first XGBoost model. Accuracy=86.75%, AUC=0.9393.
XGBoost · TF-IDF · Acc=86.75% · AUC=0.9393
PythonNLPXGBoostTF-IDFStreamlit
Live
Regression · Workplace Wellbeing
NeuroWell — Burnout Rate Predictor
Predict employee burnout rate before productivity and health collapse. Early warning over lagging consequence. R²=0.855, RMSE=0.072.
Gradient Boosting · R²=0.855 · RMSE=0.072
PythonGradient BoostingScikit-learnStreamlit
Live
Multi-class Classification · Logistics
SwiftChain Delivery Delay Prediction
Predict delivery delay category before dispatch. Notable: transparently documents a data leakage fix — because honest engineering > clean-looking numbers. Acc=62%, F1=0.579.
Gradient Boosting · Multiclass · Acc=62% · F1=0.579
PythonGradient BoostingScikit-learnStreamlit
Live
Binary Classification · Financial Inclusion
Income Level Prediction
Predict income bracket to power financial inclusion and credit scoring for underbanked populations. UCI Census dataset. 5-model comparison with SMOTE balancing.
Random Forest · SMOTE · 5-Model Comparison · UCI Census
PythonRandom ForestSMOTEStreamlit
Live
EdTech · Binary Classification · Early Intervention
Student At-Risk Predictor
Identify students likely to fail before exam season so a teacher can intervene in week 4, not after results. Built on synthetic Nigerian secondary school data. SHAP explains each prediction. Single student + CSV batch modes.
Random Forest · SHAP · Synthetic Nigerian Data · Two Modes
PythonScikit-learnSHAPStreamlit
Live
EdTech · Data Analytics Dashboard
Student Performance Tracker v3
Give teachers a real data picture of their class. 4 interactive tabs: Class Overview, Student Profiles, At-Risk Thresholds, Topic & Question Analysis. CSV-driven. No guesswork.
Streamlit · Plotly · 4 Tabs · Topic + Question Miss Rate
PythonPandasPlotlyStreamlit
Live
EdTech · Rule-Based AI · WAEC Prep
Student Study Plan Generator
Generate personalised revision schedules based on a student's weak topics, available time, and upcoming exam date. Built for WAEC and NECO prep where study time is often unstructured.
Rule-Based AI · WAEC / NECO Prep · Personalised Learning
PythonStreamlitPandas
Live
EdTech · Full-Stack Tool · SQLite
CBT Question Bank Manager
Build, organise, tag, and reuse CBT question sets across subjects and exam years. No more lost or disorganised question banks before exam season. CBT Pro's companion tool.
Streamlit · SQLite · CBT Companion Tool
PythonStreamlitSQLite
"Every project on this page started not with a dataset — but with a problem I watched real students, teachers, or organisations struggle with. Data science and AI are the tools. The classroom, the boardroom, the clinic — these are the briefs. An AI-Augmented Solutions Developer does not wait for the perfect environment or the perfect technical background. They use what they understand clearly, leverage AI confidently, and build the solution the problem actually needs."
— Adewale Samson Adeagbo · Data Scientist · STEM Educator (15+ Years) · AI-Augmented Solutions Developer