Paddy

@anime_daily before AUI and OH we had rock lee 🥀

This project transforms raw survey data from the Ghana Tech Ecosystem Survey 2025 into an actionable dashboard that provides deep insights into Ghana's technology sector. The survey captured responses from tech professionals, covering demographics, skills, tools, work arrangement

Final-year student at University of Ghana 🇬🇭 Learning that “perfect” models fail in production. Sharing wins & mistakes as I learn. Follow if you’re on the same path 👋

A 99% accuracy score once made me feel unstoppable Now I know in ML, 99% often screams overfitting Here’s why learning noise beats you in the real world

What helps? Regularization A penalty for unnecessary complexity. Focus on signal > noise. Simple models survive.

That’s overfitting. The model memorizes training data (noise + outliers). New data shows up and it panics

Think of an athlete who only trains on a flat indoor track 🏃‍♂️ Crushes practice. Falls apart on real terrain. They learned the track, not running.