Data Analyst Portfolio

About Me

Hello!

I'm a passionate Data Scientist and lifelong learner with a strong interest in machine learning, predictive analytics, and AI-driven data processing. With a background in electronics and biomedical engineering, I specialize in developing intelligent models for complex data, including brain signal analysis. I enjoy building innovative solutions that merge cutting-edge technology with real-world impact.

What I Do

I specialize in developing feature extraction methodologies. I have contributed valuable insights into analyzing brain signals and other healthcare-related datasets. Leveraging this experience, I have successfully implemented machine learning models based on the developed feature extraction methods to enhance classification and prediction capabilities.

I am available for short-term projects and part-time, longer-term engagements.

Skills & Tools

Python (Numpy, Pandas)
SQL & Databases
Data Engineering (Missing Data Handling with Imputation & Features Calculation)
Data Visualization (Tableau, Matplotlib & Seaborn)
MATLAB for Time Series Data Analysis
R programming
Statistical Analysis & Machine Learning

Projects

These are my sample works, and more on the GitHub page.

Project 1: Predicting Hospital Readmission

The prediction of hospital re-admission based on electronic health records using different machine learning techniques implemented in Python programming.

Project 2: ECG Signal Arrhythmia Detection

The automatic detection of the ECG Arrhythmia system was implemented using three different feature extraction methods such as statistical features, FFT, and CWT, and these features are classified using different machine learning techniques, including SVM, KNN, and RF.

Project 3: Prediction of The Bitcoin Prices

This project analyzes Bitcoin price trends using exploratory data analysis (EDA) and machine learning. It covers data cleaning, visualizing historical trends, computing moving averages, and building a simple predictive model with linear regression.

Project 4: Wearable Sensor Data Analysis

This project analyzes wearable sensor ECG data to extract time-domain and frequency-domain features, such as heart rate and power spectral density, for classifying different physical activities and visualizing the results using dynamic plots and a dashboard on Tableau.

Project 5: Heart Rate Variability (HRV) Analysis

This project provides an interactive tool to analyze Heart Rate Variability (HRV) by visualizing HR and RMSSD trends over time, helping users understand stress, recovery, and cardiovascular health through condition-based insights. The app is built using Streamlit to create an interactive web interface.

Papers

These are my published papers and more on Google Scholar.

Modified-Distribution Entropy as the Features for the Detection of Epileptic Seizures

The proposed modified-Distribution entropy (mDistEn), combining advantages of fuzzy and distribution entropy, achieves 92% classification accuracy and higher AUC values for epilepsy detection in EEG data compared to state-of-the-art entropy methods.

Prediction of epileptic seizures based on multivariate multiscale modified-distribution entropy

The proposed multivariate multiscale modified-distribution entropy (MM-mDistEn) combined with an artificial neural network (ANN) achieves 98.66% accuracy, 0.84 AUC, a 0.014/h false alarm rate, and 26.73-minute average prediction time for epilepsy seizure prediction, demonstrating strong clinical potential.

Entropy-Based Emotion Recognition from Multichannel EEG Signals Using Artificial Neural Network

TThe proposed multivariate multiscale modified-distribution entropy (MM-mDistEn) combined with an artificial neural network (ANN) achieves high emotion recognition accuracy from EEG data, with average accuracies of 95.73% ± 0.67 (valence) and 96.78% ± 0.25 (arousal) on the GAMEEMO dataset and 92.57% ± 1.51 (valence) and 80.23% ± 1.83 (arousal) on the DEAP dataset.

Si Thu Aung