codewars-python

coursera-databases-and-sql-for-data-science-with-python

coursera-programming-for-everybody

coursera-python-data-structures

coursera-using-python-to-access-web-data

crowdeye

gdsc-ai-stock

leetcode-python

ml-forest-fire-prediction-with-regression-and-classification

Using both Regression and Classification to make a good prediction to an extreme imbalance dataset. Improve the R square from 0.01 to 0.73 for Regression and the Accuracy of Classification can up to almost 90%.

ml-mathematical-and-statistical-foundation-of-shrinkage-method

In-depth exploration of the mathematical and statistical principles behind the shrinkage method. Using Lasso and Stepwise feature selection to select and classify an artificially expanded variable data set.

ml-wine-type-and-quality-classification

In wine type classification, seven distinct supervised learning classification models were evaluated. XGBoost emerged as the frontrunner with an impressive accuracy rate of 99.46%. However, when it came to the classification of wine quality, our approach diverged from established literature, presenting alternative perspectives and viewpoints.

nlp-sentiment-classifier

Utilize and explore the mathematical foundation of Generative and Discriminative algorithms to build a Natural Language sentiment classifier with Twitter samples dataset.

nlp_dl-topic-modeling-for-tvl-livestream-comments

這是一份和企業甲級排球聯賽（以下簡稱企排）合作的研究，旨在為企排 YouTube 直播留言建立一個主題模型以量化分析觀眾的討論話題，進而達到了解各主題隨時間的熱度 分佈及掌握觀眾的注意力，以利後續開發更多的商業用途。本文所使用的所有資料均源自於 企排 18 年所有直播場次的留言資料，資料分別透過五個預處理方式以評估模型表現。分類模型由三個分類器所構成，分別用來分類主要主題（閒聊、比賽、加油、轉播）、次要主題（將比賽細分為球員、球隊、裁判、教練、戰術）以及情緒分析，三者的量化評估分數 Area Under ROC Curve 高達 99.55 / 99.73 / 99.99，單句留言平均計算時間（計算於 Nvidia T4 GPU）為 0.044 seconds / sentence。

random-walk-distribution

scfengv

Hi there, check out some works I have done in my spare time

ship-analysis

stock-valuation

This project serves as a 5-year DCF model to evaluate companies' target prices, using Selenium to fetch stock information mainly from Yahoo Finance & Stock Analysis. The information includes everything needed to calculate DCF.