Kaggle & Datathons: A Practical Guide to AI Competitions
- 1. Kaggle & Datathons A Practical Guide to AI Competitions On Campus: Birzeit University
- 2. Editable University Name What is Kaggle? What is a Datathon? Break☕ Live Datathon Demo Q&A Agenda
- 4. ● “The Home of Data Science” ● It was founded in 2010 and acquired by Google in 2017 ● There are 22,538,110 kaggle users, There have been 5,716 competitions held, There are 436,580 public datasets, There have been 443,454 topics posted, with 2,623,311 comments made Kaggle
- 5. Key Features Competitions Grow your data science skills by competing in our exciting competitions. Datasets Explore, analyze, and share quality data. Code Explore and run machine learning code with Kaggle Notebooks. Discussions Discuss the Kaggle platform & machine learning topics – this includes sharing feedback, asking questions, and more. Learn Gain the skills you need to do independent data science projects.
- 6. What is a Datathon? Chapter Two
- 7. Competition on anything is good, because it makes everybody better. On Campus: Birzeit University
- 8. A Datathon is a data science competition where participants are given a dataset and a problem statement. The goal is to analyze the data and develop predictive models or insights, typically within a limited timeframe. Datathon ● They simulate real-world data science problems. ● Companies use Datathons in job interviews for AI/ML roles. ● They provide practical experience beyond what courses offer. ● You build a portfolio of projects, which can help in job applications.
- 9. Datathon Flow chart Start Registration & Team Formation Receiving the Dataset & Problem Statement Data Exploration & Preprocessing Feature Engineering & Model Building Submission & Evaluation Closing & Prizes End
- 10. Read it twice and note the evaluation metric, Check if the submission format is clear ● Understand the Problem Statement Garbage in = Garbage out. Missing values? Outliers? Duplicate entries? Visualizing the data helps identify trends and errors. ● Focus on Data Cleaning & Exploration Baseline models (Linear Regression, Decision Trees) give quick insights, Avoid getting stuck optimizing a complex deep learning model too early ● Start with a Simple Model Kaggle Discussions and Winning solutions are often public learn from others. ● Learn from Past Solutions Strategies for Success in Datathons 🏆
- 11. ☕ BREAK
- 13. ❓ Q&A
- 14. Thank You
Editor's Notes
- #4: https://www.kaggle.com/code/carlmcbrideellis/kaggle-in-numbers