An implementation of Gradient Boosting Decision Tree (GBDT) for binary/multiclass classification from scratch. Include an improved base classifier
CART which can handle missing value. Techniques: Python
Implemented and analyzed four clustering algorithms for mixed type data on six datasets, including KMCMD, SBAC, Frequency Neuron Mixed Self-Organizing Map, and AUTOCLASS.
Techniques: Python, R
Developed a prediction model using Markov Chain and Recurrent Neural Network (RNN) to forecast gold and bitcoin prices. Implemented a robust trading
strategy based on the predicted prices for efficient buying and selling decisions. Techniques: Python, scikit-learn
Designed and implemented a simulation model to predict the number of tries of the Wordle Game, incorporating dictionary partitioning strategies and human decision behavior. Applied
ridge regression to determine the relationship between selected factors and the level of difficulty of a given the word.
Techniques: Python, scikit-learn
Developed a Lethal Collision Probability Risk Model based on average lethal probabilities within different regions. Processed spatial data and employed visualization techniques to identify high-risk areas
and slow zones for enhanced risk assessment and mitigation. Techniques: Python
Developed a mini programming language, incorporating essential features such as Booleans, Naturals, Simply Typed Pure λ-Calculus, Let Bindings, Sequencing Operator, and more.
Implemented the semantics of these language features and performed type check. Language: Haskell
Developed a 3D snake game utilizing OpenGL. Implemented advanced features
including first-person perspective, skybox, keyboard control, and realistic lighting to enhance the gaming experience. Language: C++