Skip to content
View noobarafat's full-sized avatar

Highlights

  • Pro

Block or report noobarafat

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
noobarafat/README.md

K. M. Arafat Islam, Explainable AI Researcher

Email GitHub followers Profile views


🧊 My GitHub, in 3D

3D contribution graph of noobarafat

Rebuilt every day from my contribution history.


About me

I build machine learning models that have to explain themselves. Most of my work pairs ensemble and gradient boosting models with SHAP and LIME, so the people making decisions in health, climate, supply chain, retail and HR can see why a model says what it says.

πŸ”¬ Researching Explainable AI (SHAP, LIME) for socioeconomic and business data
🏒 Building Buildsign Digital, a digital agency (Founder and CEO)
πŸ’Ό Working at Ostad
πŸŽ“ Studied B.Sc. in Computer Science and Engineering, Daffodil International University
🌱 Next Funded master's study in Data Science and Business Analytics

πŸ“„ Research in 2026

Venue Paper Status
IEEE CSDE 2026 Ensemble-Based Detection of Global Climate Anomalies Using Surface Temperature and COβ‚‚ Emission Indicators: A Majority-Voting Unsupervised Framework with SHAP Interpretability βœ… Accepted
IEEE CSDE 2026 Global Diabetes Mortality Analysis: A Reproducible WHO Data-Driven Study of Trends, Regional Disparities, and Prevalence-Mortality Relationships (2000–2014) βœ… Accepted
IEEE i-COSTE 2026 Sector-Level Emission Intensity Profiling of Supply Chains: A Leakage-Free Machine Learning Approach Using Structurally Independent Predictors βœ… Accepted
IEEE WIECON-ECE 2026 UERAF: A Unified Framework for E-Commerce Return Prediction, Root Cause Analysis, and Revenue Impact Simulation βœ… Accepted
COMPAS 2026 Targeting the Persuadable: A Causal Uplift Framework for Budget-Constrained Retail Campaign Optimization βœ… Accepted
iCACCESS 2026 An Interpretable Ensemble Learning Framework for Predicting Students' Entrepreneurial Intention in Digital Entrepreneurship Education 🎀 Presented

⭐ Featured project

Research Paper Humanizer

Research Paper Humanizer is an evidence-based Claude skill that rewrites AI-assisted research papers so they read as author-written. It lowers Turnitin, GPTZero and Originality.ai scores while keeping every number, citation and technical term unchanged. It is built on peer-reviewed studies of how LLMs write science.

Stars

🧰 Toolkit

Machine learning and XAI

Python scikit-learn XGBoost LightGBM CatBoost SHAP LIME Optuna

Data and analysis

pandas NumPy SciPy Matplotlib SQL Power BI Excel Google Colab Jupyter

Design and build

Figma UI/UX Design Mobile app development Git Claude

πŸ† Highlights

πŸ“‘
Featured by
Grameenphone
πŸ“±
Govt.-certified
mobile app developer (EDGE)
🎨
Fiverr Level 1
UI/UX designer
🀝
100+ events
organized and volunteered

Open to research collaborations in explainable AI, applied ML and business analytics.
Get in touch

Pinned Loading

  1. research-paper-humanizer research-paper-humanizer Public

    Evidence-based Claude skill to humanize AI-assisted research papers and lower Turnitin, GPTZero and Originality.ai AI-detection scores without changing meaning, citations or academic quality.

    4