Ornrakorn's Portfolio Website

Programming Projects

Most recent work

Photo by Robina Weermeijer on Unsplash

Heart Failure Prediction

Heart failure prediction involves using data to forecast whether an individual has normal or heart failure. The dataset for this task is imbalanced. The results indicate that Random Forest is the most effective model, achieving an accuracy of 90% and an AUC value of 94%.

Photo by KOMMERS on Unsplash

Stroke Prediction

Stroke prediction involves utilizing data to predict whether an individual's condition is normal or stroke. The dataset for this task is highly imbalanced. I have experimented with various techniques and multiple models in this project to address the challenge posed by imbalanced data.

Pneumonia Detection using CNN

Pneumonia detection using CNN involves predicting whether a given Chest X-ray image indicates a normal condition or pneumonia. The optimized CNN model, equipped with class weights and augmented data, achieved outstanding performance, boasting an accuracy of 92.7% and an impressive F1-Score of 94%.

Denoising Dirty Documents with AutoEncoder

The use of AutoEncoder, which is a type of neural network and an Unsupervised learning approach, assists in removing noise from documents. For example, wrinkles, coffee stains, and shadows from document scans allow documents to become more readable than before.

STEHealth Application

STEHealth is a shiny application for analyzing space-time patterns, cluster detection, and association with risk factors of health outcomes.

Spatiotemporal epidemiology and analysis of mental health conditions in Thailand project

Analyzed mental health service utilization in Thailand using spatial pattern analysis and cluster analysis.

Photo by EKATERINA BOLOVTSOVA on Pexels

Churn Classification

Created churn classification models including Decision Tree and Random Forest as part of the Intermediate Data Science (2nd-gen) by Thammasat University Research and Consultancy Institute (TU-RAC) workshop.

Sale Report Dashboard

Created a sale report dashboard as part of the Intermediate Data Science (2nd-gen) by Thammasat University Research and Consultancy Institute (TU-RAC) workshop.

Analysis of mental health conditions in Thailand Dashboard

This dashboard was created after the Spatiotemporal Epidemiology and Analysis of Mental Health Conditions in Thailand project, which was conducted to train myself in using Tableau.

Photo by Karolina Grabowska on Pexels

Cardiovascular Disease Prediction Project

Studied and compared algorithms used to accurately diagnose cardiovascular disease, which is a disease with high mortality statistics.

Online Retail Dashboard

Created a dashboard for online retail during the FutureSkill workshop on the analyzing data with Power BI.

Topic Modeling From Personalized Medicine: Redefining Cancer Treatment

Investigated the meanings of each class (all 9 classes) in this dataset and utilized Topic Modeling as a helpful technique.

Photo by Snapwire on Pexels

Dog or Cat (Convolutional Neural Network Lab)

Built a Cat vs Dog classifier model as part of the CPE378 Machine Learning course.

Handwritten Digit Recognition (Convolutional Neural Network Lab)

Built a deep convolutional neural network model for handwritten digit recognition using the MNIST dataset of handwritten digits as part of the CPE378 Machine Learning course.

Photo by Emojisprout emojisprout.com on Unsplash

Emoji Prediction (Recurrent Neural Network Lab)

Predicted the emoji that represented the emotion in each sentence as part of the CPE378 Machine Learning course.

Photo by Ali Hajiluyi on Unsplash

Heart Disease Prediction Lab

Built a decision tree model to classify patients with heart disease as part of the CPE352 Data Science course.

Photo by Andrea Piacquadio on Pexels

Telco Customer Churn Prediction Lab

Built a decision tree model and a linear discriminant analysis (LDA) model for customer churn prediction using the Telco dataset as part of the CPE352 Data Science course.

Landing page website for a product or service

Created a landing page website for a product or service during the FutureSkill workshop on the fundamentals of HTML5, CSS3, and JavaScript.