Welcome

Hi, I’m Wasik Billah. I am an Electrical and Electronic Engineering graduate from the Islamic University of Technology (IUT), Bangladesh.

My Works

• Load forecasting and scheduling: I benchmarked seven models (XGBoost, LightGBM, Random Forest, Gradient Boosting, KNN, SVR, MLP) on hourly feeder data from 10 cities in the Rangpur division. XGBoost performed best, with an R² of 0.944 and a MAPE of 2.69%. I used K-means clustering on the forecast to split the day into peak and off-peak hours. Hybrid metaheuristic schedulers HGWOCSO then moved household appliances to off-peak hours. In my papers, these schedulers reduced the peak-to-average ratio by up to 52.3% and electricity cost by up to 15.6%.
• Medical data: A comparison of classifiers on demographic and lab data to detect Hepatitis C, where AdaBoost came out most effective.
• Image classification: A seagrass classifier built on a frozen pretrained VGG19 base, with Batch Normalization, Dropout and a softmax layer on top.
• Battery health estimation: A machine learning pipeline that estimates lithium-ion battery State-of-Health from NASA discharge data (34 cells). It uses physics-based features, and the best model, Random Forest, reached an RMSE of 0.086.

How I work

I like to break down a problem into small parts and solve them step by step. I like to visualise the whole pipeline. I am also very punctilious when it comes to taking notes (I use Obsidian extensively).

Publications

Education

Secondary Education ▸

Experience

Projects

Explore my technical projects on GitHub.

Membership

Extracurriculars ▸