Michael Ye's Machine Learning Porfolio
Welcome to my portfolio. I'm a self-taught Machine Learning Engineer who has built a foundation for Machine Learning through complex online self-enrolled college-level courses. And now, I have enough experience to start creating a robust collection of competitions and projects.
Feel free to look at my projects (and their corresponding websites where you can try my deployed models), my courses, and more!
My Projects!
My Best Score:
I achieved a high score of 80.622% in this project and top 587 out of 67,149 Entries (top 0.8%) and counting. Here's a snapshot of my score:
Titanic Survival Machine Learning
This is my first Machine Learning Project. My GitHub repository contains my work for the Kaggle Titanic survival prediction competition. The challenge is to use machine learning to create a model that predicts which passengers survived the Titanic shipwreck based on a range of features.
The project employs a variety of machine learning models, including Random Forests, Gradient Boosted Trees, and Neural Networks, to predict survival. Techniques for data cleaning, feature engineering, and model tuning are thoroughly documented in the Jupyter notebooks.
I also made a custom interactive website for this project, which describes the project, my techniques, and most importantly, I deployed my best models and allowed users to pass in values which are given to the model in the cloud and gives back predictions of survival to the user.
Histopathologic Cancer Detection Project
This is my second Machine Learning Project. My GitHub Repository displays all my code. This project is based on a Kaggle competition with the goal of developing an image classification algorithm using Convolutional Neural Networks (CNNs) to identify metastatic tissue in histopathologic scans of lymph node sections.
The primary challenge in this competition is to create a model that can accurately classify microscopic images of lymph node sections as containing metastatic tissue or not. This repository includes the code for building and training a CNN to tackle this problem. The model is developed in Python using TensorFlow and Keras.
I also made a custom interactive website for this project for this project. In this website, I show how my project leverages a Convolutional Neural Network (CNN) model to analyze pathology scans uploaded by users and predict the presence of cancerous cells. The application is developed using Streamlit and TensorFlow and is part of a Kaggle competition aimed at advancing cancer detection methods.
Nepal 2015 Earthquake Building Damage Predictor
This is part of the "Richter's Predictor: Modeling Earthquake Damage" competition hosted by DrivenData. This is my third Machine Learning Project. The project aims to predict the level of damage to buildings caused by the 2015 Gorkha earthquake in Nepal based on aspects like building structure and location. This model is an essential tool for disaster management and helps in prioritizing aid and reconstruction efforts.
I achieved a high score of 74.82% in this project. The #1 global score is currently 75.58%. I'm currently in the top 420 out of 7,221 (top 5.8%).
And just like the other projects, I made a custom website for this project, where I got into a lot more depth about what I learned, my thought process for each model that I used, the hyperparameter tuning techniques I employed, and best of all, you get to try ALL my models that are deployed from the cloud. Pass your own values into the model (using my interactive website) and then get back a prediction on the level of damage your building would have experienced in the earthquake.
The Most Important Courses I've taken:
Andrew Ng's Machine Learning Specialization
I completed Andrew Ng's renowned Machine Learning Specialization Course, an immersive three-part series encompassing 'Supervised Machine Learning,' 'Advanced Learning Algorithms,' and 'Unsupervised Learning.' This comprehensive specialization deepened my understanding of the core principles of machine learning. I gained proficiency in supervised techniques, tackling problems where guidance through labeled data is available. The course on advanced algorithms enhanced my skills in complex concepts like neural networks and deep learning. In the unsupervised learning segment, I explored data-driven decision-making in scenarios lacking explicit guidance, mastering clustering and dimensionality reduction techniques. This specialization was pivotal in refining my machine learning expertise, equipping me with both theoretical knowledge and practical skills to address a wide array of real-world challenges.
Harvard's CS50 Introduction to Computer Science
I completed Harvard's CS50, an intensive dive into the fundamentals of computer science. This course honed my problem-solving skills and introduced me to the principles of software engineering. I gained hands-on experience in a variety of languages including C, Python, and JavaScript, and explored the intricate world of algorithms and data structures. The rigorous problem sets challenged me to think algorithmically and taught me best practices in code efficiency and readability. This course laid a robust foundation for my subsequent forays into the field of Machine Learning.
Harvard's CS50's Introduction to Artificial Intelligence with Python Science
I'm proud to have completed Harvard's CS50AI: Introduction to Artificial Intelligence with Python, where I deepened my understanding of AI principles and applied them using Python. The course was a journey through the landscape of algorithms and machine learning, covering topics from knowledge representation to neural networks. I learned how to design intelligent systems that can make decisions, learn from data, and exhibit behavior akin to human intelligence. This course has been pivotal in equipping me with the skills to build sophisticated AI models and has significantly contributed to my expertise in the field.
MIT's 6.00.1x: Introduction to Computer Science and Programming Using Python
Completing MIT's 6.00.1x: Introduction to Computer Science and Programming Using Python was a transformative experience in my coding journey. The course provided me with a solid grounding in computer science concepts and practical Python programming skills. I learned computational thinking and problem-solving techniques, enabling me to write efficient algorithms and process large data sets with Python. The hands-on problem sets and projects were crucial in developing my ability to apply theoretical knowledge to real-world challenges, laying the foundation for my future work in Machine Learning and AI.
Harvard CS50’s Introduction to Programming with Python
Harvard CS50’s Introduction to Programming with Python marked a significant milestone in my programming journey. In this course, I delved into the fundamentals of Python, mastering its versatile features and capabilities. I learned to think algorithmically, developing skills in writing clean, efficient code and solving complex problems effectively. This course broadened my understanding of programming concepts, data structures, and algorithms, providing a solid foundation for my future projects and endeavors in the dynamic field of computer science
