• Name: Seoyun Kim
• Email: seoyun.kim@g.skku.edu
• Interests: ML, DL, Applied AI

I am Seoyun Kim, a 2nd year graduate student who is double majored in Data Science & Artificial Intelligence at Sungkyunkwan University, Seoul, Korea. I am currently I am a M.Sc. student in Data eXperience Lab at Sungkyunkwan University, advised by Eunil Park.

As a member of the categorical data analysis team of P-Sat of the Statistical project team of Sungkyunkwan University, I have experience developing various models and handling categorical data. In addition, various projects related to Data Visualization, Prediction ML Model, DL Model such as LSTM, NLP techniques, and Data Preprocessing have been carried out. I also participated in various data analysis/AI researches in DX Lab, including researches using image, text, video, and time series data.

• Data Analysis
• Computer Vision
• Machine Learning and Deep Learning
• Artificial Intelligence(AI)
• Multimodal Modeling
• Social&Affective Computing

• D-ViSA: A Dataset for Detecting Visual Sentiment from Art Images pdf github

  • In Proceedings of the IEEE/CVF International Conference on Computer Vision
  • Built abstract art image dataset annotated with dimensional emotion labels, conducting deep learning model experiment for detecting dimensional emotion from art images

• Understanding mental health issues in different subdomains in social networking services: computational analysis of text-based reddit posts github

  • Journal of medical Internet research
  • Examined and classified the linguistic characteristics of user posts on specific mental disorder subreddit channels (depression, anxiety, bipolar, borderline personality disorder, schizophrenia, autism, and mental health) on Reddit using sentiment analysis and unsupervised clustering methods

• Micro-Locational Fine Dust Prediction Utilizing Machine Learning and Deep Learning Models github

  • Computer Systems Science and Engineering
  • Predicted micro-locational fine dust concentration from air quality and meteorologucal time-series data using ML/DL models

• M.Sc. in Data Experience Lab at SKKU (2022.02. ~ present)

• A Member of P-Sat at SKKU (2020.08. ~ 2021.02.)

• Data Science Team Manager at Dacon (2021.12. ~ 2022.02.)

• Dam Water Level LSTM Prediction link
  Predicted and Analyzed by using statistical methods and deep leaning RNN model - LSTM - to predict water-level of the dam using multi-variable dataset
  

• Real-time news crawling link
  Built real-time news crawling engine including search keyword by using BeautifulSoup and made news data preprocessing module
  

• Binalry Classification in Predicting Political Party link
  During 'Theme Analysis' we used statistical metodes including t-test, homogeneous test and variable selection, EDA & feature engineering for Preprocessing
  Modeling using Ensemble model, XGBoost, Light GBM, and Random Forest in order to predict the 'Party' variable
  
  Building prediction model for Inbalanced dataset using PCA, SMOTE, and various Prediction model such as Ligh GBM and Cross Validation and measured F-1 score in 'Kaggle competition'
  

• Predicting Wheter-to-vote link
  Predicted wheter the person will vote or not using psychological survey dataset with XGBoost model
  

• Wine Filtering and Recommendation System link
  Built wine recommendation program using QtPy and filtering methods
  

• Visualization of Alchol Cunsumption around the Globe link
  Using R, visualized the correlance between happiness score, region, and alcohol consumption