Hi, I'm Run and I'm a data scientist and computer vision. I have developed and launched several machine learning and NLP products, including an AI cardiologist, a deep learning framework, and a job recommendation tool. I have gained experience in various industries, including in the medical, education, and insurance sectors, and recently graduated with first honors. I am now ready for full-time work and welcome the opportunity to work with you. Please feel free to contact me and check out my portfolio and previous projects to learn more about my skills and experience. I also run a YouTube channel and Facebook page where I share my learning journey and tips for data scientists and software engineers.
Contact me : runpan4work@gmail.com
Last updated 25 Dec 2022
a Python package that simplifies the process of constructing neural network architectures.
It is especially useful for beginner data scientists who want to create logistic, shallow,
or deep L-layer neural networks with just a few lines of code. The author of Dipple developed
the entire neural network architecture from the ground up using NumPy, a linear algebra library.
Hobot is a cardiologist trained to detect early signs of heart failure in people with a history of heart failure and left ventricular systolic dysfunction. Hobot uses a Random Forest Classifier with 89 trees and 12 features (eg. CPK enzyme) to predict the probability of heart failure. The web application featuring Hobot includes a museum section displaying the training dataset and providing educational insights through statistical analysis.
Read More
Need help finding the perfect emoji for your English text? Try COMMUJI,
Our tool uses advanced NLP technology, specifically BERT-base-uncased,
to suggest the most appropriate emojis
with a weighted F1 score of 0.9247. Say goodbye to confusing your friends or colleagues
with the wrong emoji.
The goal of this project is to forecast consumer interest in sunscreen in Thailand by analyzing Google search data for "ครีมกันแดด", The research used various models and found that that the SARIMA(0,1,2)(0,1,1)(12), which we have named Sunrima, was the best fit based on the AIC criterion. This information can be used by companies selling sun protection to determine the best time to launch targeted advertisements and marketing campaigns for sunscreen to increase product sales and revenue.
Read MoreCareerMatcher helps you find the best job for your unique characteristics near real-time by using BeautifulSoup to scrape relevant job listings based on your job title or company name from JobsDB. Simply input your characteristics, such as strong communication skills and Python and Tableau, and CareerMatcher will use TFIDF vectorization to rank the jobs that best match your profile. With CareerMatcher, you can easily find the job that aligns with your strengths and values.
Read MoreWhen traveling, you may have limited space in your bag for food. As a healthy individual, you need to ensure that the food you choose has sufficient dietary fiber. Some high-fiber foods may contain high amounts of sugar, which can be harmful. To maximize dietary fiber benefits and minimize sugar's negative effects, it is important to carefully select your food. This can be time-consuming, but our fast and smart food picker, based on genetic algorithms, can help you choose the best options based on your criteria.
Read More
Hidden Markov Models (HMMs) are commonly used for Part-of-Speech (POS)
Tagging in Natural Language Processing. For this task, we will focus on
a simplified version of POS tagging where only nouns, verbs, and modifiers
This will demonstrate the basic concepts of HMMs for POS tagging from scratch in Python.
When traveling, you may have limited space in your bag for food. As a healthy individual, you need to ensure that the food you choose has sufficient dietary fiber. Some high-fiber foods may contain high amounts of sugar, which can be harmful. To maximize dietary fiber benefits and minimize sugar's negative effects, it is important to carefully select your food. This can be time-consuming, but our fast and smart food picker, based on genetic algorithms, can help you choose the best options based on your criteria.
Read More
We are opening a new bubble tea shop and have three options on our menu:
milk tea, green tea, and the choice to add whip cream or not.
We also have the option to add brown sugar or not. We asked our friends to rate
the tastiness and appearance of each combination of these options via Google form,
but the ratings are varied. To understand the impact of each factor and potentially
the interaction between factors, we are using a 2^3 full factorial design.
As a parent, you may want to restrict certain songs to avoid inappropriate content
for your child. With the YouTube Music PC App, you can add metadata for approved songs
to a SQL database. The app's GUI, created using the Tkinter package, will only display
the songs that have been approved by the parent. The app also stores song metadata in a
SQL database using the sqlite3 package
Last updated 25 Dec 2022
Highlighted Courses : Statistical Data Analysis, Database Management, Operations Research, Simulation Modelling, Calculus of Several Variables, Advanced Calculus, Operations Research, Numerical Analysis, Data Structure in Mathematics, Linear Algebra, Supply Chain Modelling & Optimization, Machine Learning, Big Data Analytics, Statistical Modelling, Stochastic Processes & Applications, Experimental Design Analysis, Data Mining, Mathematics For AI, Mathematical Statistics, Intro to Mathematical Software (MATLAB), Network Optimization
View my student transcriptLast updated 25 Dec 2022
As a main leader in the development of a natural language processing (NLP) model for fraud detection in the insurance industry, my work involves using advanced natural language processing techniques to identify fraudulent interactions between agents and customers. My works also involves interacting with API servers to ensure that the model is integrated into the broader system and can be accessed as needed
As a data analyst at TDA, my work involves an automation system that transforms
raw data into clear and concise data analysis reports and dashboards. I carefully
select the most relevant and crucial plots to present to our audiences.
This system saves time and resources while producing high-quality results.
As a Data Science Intern, I lead discussions on the use of natural language processing (NLP) models for sepsis detection and predicting pain levels in cancer patients. I analyze the performance of traditional and machine learning approaches and recommend which model to implement in the future. I also extract data from doctor prescriptions using regular expressions and turn it into usable features for NLP tasks.
Last updated 25 Dec 2022
I participated in a competition with the Wanchem team and we were awarded the popular vote prize. Our project involved using machine learning techniques, such as encoding the names of molecules into 2D matrices and using a convolutional neural network (CNN), to predict the gas adsorption ability of metal-organic frameworks (MOFs). Our model had a mean absolute error loss of 1.245, which placed us at rank 26 out of 250 teams.
Read More
Launch Online Activities to optimizing productivity by applying Physical Therapy Knowledge,
held on Kanjanapisek Witthayalai Nakornphathom School.
Last updated 25 Dec 2022