Emmanuel | Data Analyst 📊📉

Hi #DataFam, meet the Personal Finance Tracker Money management today is messy — cards, bills, savings and investments all over the place Spreadsheets? Too static So I built a product that tracks money & builds financial discipline Explore it here: public.tableau.com/app/profile/fa…

- Real dropout rate shifted from 12.8% to 8.2% The model got more honest and it was worth every time. Updated Github documentation: github.com/ENGFAKAYODE/St… Grateful to @Excelerate_org for the opportunity to grow and see firsthand how AI can power data analysis

When I engineered a new feature (Has_Start_Date): - Model recall jumped from 39% to 79% after SMOTE rebalancing - 55 additional at-risk learners flagged per cycle who would have received no intervention - Learner sign-up volume forecasted with confidence intervals

I did not just complete the internship. I was named a Star Performer. I joined the @Excelerate_org AI-Powered Data Analysis Remote Internship, aimed at boosting student engagement and reducing drop-offs of Learners on Saint Louis University Opportunity Wise Platform.

Then they handed that clarity to AI and let it move faster That is the sequence most people are skipping Data first Clarity second AI third Not the other way around So before you invest in AI tools, answer this Do you actually know what your data is telling you right now?

The founders winning with AI right now are not the ones who adopted it first They are the ones who understood their business through data first They knew their numbers They knew what was driving growth and what was draining it

AI will not save your business. At least not until you fix this first Every week a founder tells me they want to use AI to grow They think of more automations, smarter decisions and faster insights But when I ask what their retention rate is they go quiet

Though accuracy dropped from 94.22% to 86.89% the trade-off is worth it. I documented the entire process, insights and recommendations here: lnkd.in/ejVjEAqu

The results: - Gradient Boosting recall jumped from 39.13% - 78.99% - The model now catches 109 at-risk learners per cycle instead of 54 - Logistic Regression & SVM went from zero F1 to functional - 55 additional real people flagged who would have received no intervention before

After Week 4 of my Excelerate ended, I did not stop working I personally applied the feedback we received during our final presentation Instead of dropping the 3,794 rows with missing Opportunity Start Dates I flagged them

@EmmanuelAanu1 @followKWSG Congratulations to you Well done

I am honored to be invited by the Kwara State Government @followKWSG to speak at the Closing Ceremony of the 3-Day Capacity Building Training for Student Entrepreneurs where I’ll be sharing my SUNPOD journey, lessons on innovation & building impact.