Hasitha Eranda

Data Scientist| Astrophysicsist

Self-driven, motivated Scientist with 8+ years of proven impactful experience in data science across physics and healthcare. Scientific experience spans in Machine Learning Models, Predictive Analytics algorithms and toolkit experience spans in Python, SQL, HPC, Linux. Enthusiastic about leveraging data science skills for developing mathematical models leading to data-driven decisions.

My Work in a Nutshell

Do Simulations produce relic galaxies?

How to Evaluate Credit Default Risk?

Can you predict Solar Activity?

Advanced Machine Learning for Predicting Hospital Readmissions in Diabetes Care

Optimizing the best location to observe a Solar Eclipse

Projects

Building a Predictor for Credit Default Risk Evaluation

Python Jupyter XGBoost, Random Forest, Incremental Learning

Feature Importance Visual — ROC of the predictor for XGBoost model

Applying ML algorithms for Time Series Modeling of Sunspots

Python Jupyter XGBoost, LSTM, PINNs

I applied various ML algorithms to predict the number of sunspots per month and Hyperparameter tuning analyzed the performances.
Whats new? Developed Novel loss functions and implemented to improve time series forecast.
The application of a new algorithm in time series forecast, Temporal Graph Neural Networks .

Sunspot Predictions by the LSTM model with a newly defined loss function.

Applying ML Models to Optimize Diabetes Treatments in Hospitals

Python Jupyter XGBoost, Random Forest, Deep Neural Networks, Logistic Regression

What's new? We applied XGBoost with incremental learning to efficiently handle large feature space and memory limitations, overcoming system crashes due to limited RAM.
We implemented Boruta algorithm for feature engineering inorder to optimize the machine lerning training process.

Search for Massive Relic Galaxies in Cosmological Simulations

Python Jupyter Robust Regression, HPC

Physics Informed Neural Networks (PINNs)
Machine Learning Applications in Healthcare
Symbolic Regression
Time Series Forecasting
Risk Prediction Models
Machine Learning in Finance

Public Talks/Presentations

What's wrong with dark energy? Public talk given to undergraduate students at University of Colombo.
Participated in Philosophy of Dark Energy Workshop at UCIrvine.
Guest Lecturer: Stedu Association Group. Conducted a lecture series for a summer school for middle school students.
Organizer of Physics Talks : A weekly meetup for undergraduate students at university of Colombo. I presented new updates in Physics and complex concepts.