Hawaii Machine Learning Meetup

Kaggle Competition

Test your skills at Hawaii’s first Machine Learning Competition. Whether you’re a machine learning expert or just want to prevent the robots from taking over, everyone is welcome.

Your challenge is to predict online retail sales from transaction data. Use any algorithm you like: train your own neural networks, grow your own decision trees, even spawn an AI to solve the problem for you.

If you are new to machine learning, come and join our weekly tutorial sessions to learn the fundamentals of machine learning. If you are are experienced with machine learning, come and show off your skills.

Join us to advance the field of machine learning in Hawaii.

If you are looking to participate in the competition please use the following link.

If you have any questions, contact us as at hawaiimachinelearning@gmail.com

Competition Details

General Objectives

The competition will provide a low-pressure environment that

  • Promotes machine learning in Hawaii
  • Provides an opportunity to gain experience analyzing data
  • Teaches fundamental machine learning techniques

This competition will be valuable to beginners looking to get a start in machine learning. At the same time, this competition will be beneficial to machine learning experts who are interested in applying start-of-the-art techniques to solve real-word problems.

Task

Participants are tasked with predicting sales transactions for online retail stores. Methods developed in this competition will be readily generalized to sales forecasting tasks here in Hawaii.

Evaluation

The competition will evaluate participants objectively using the data rather than subjectively based on human judgement. Participants are ranked on how well they can predict test data (the ground truth is withheld from the participants). The metric that we will use for ranking is the root mean squared log error which is standard for this type of problem.

Rules

  • Participants can form teams of any size.
  • Private sharing of code or data outside of teams in not permitted. It’s okay to share code if made available to all participants on the forums.
  • Having multiple accounts per participant is not allowed.
  • Participants can submit up to 5 submissions per day. Participants will be allowed to submit 2 submissions for evaluation on the test set.

Tutorials

We plan to hold optional tutorials to teach the fundamentals of machine learning. The tutorials are created by machine learning experts. Tutorials will be given in-person, but we will also try to stream the content.

  • Tutorial 1: Exploratory Data Analysis (EDA) and Feature Engineering
  • Tutorial 2: Training and Prediction
  • Tutorial 3: Cross Validation and Model Selection
  • Tutorial 4: Model Ensembling and Blending

Kernels

We plan to have a panel of judges who subjectively evaluate kernel submissions based on the following criteria:

  • Originality – does the kernel contain original material
  • Helpfulness – does the kernel provide valuable information for the task at hand; the number of upvotes is one measure of helpfulness
  • Correctness – are the technical details of the kernel correct, this is the most important aspect of the kernel

The best kernel will be recognized.

Competition Timeline

The competition is planned to start on March 30th and to finish on April 30th. We will wrap-up the competition on May 4 th.

Week 0, Friday March 30, 6:00 PM: Start Competition

  • Description: Kick-off the event. Discuss the competition task, rules, and timeline. Setup complete beginners. Encourage team formation.
  • Venue: Presentation Room at Manoa Innovation Center

Week 1, Monday April 2, 6:00 PM: Tutorial 1

  • Description: Exploratory Data Analysis (EDA) and Feature Engineering
  • Venue: Ideation Room at the Manoa Innovation Center

Week 2, Monday April 9, 6:00 PM: Tutorial 2

  • Description: Training and Prediction
  • Venue: Ideation Room at the Manoa Innovation Center

Week 3, Monday April 16, 6:00 PM: Tutorial 3

  • Description: Cross Validation and Model Selection
  • Venue: Ideation Room at the Manoa Innovation Center

Week 4, Monday April 23, 6:00 PM: Tutorial 4

  • Description: Model Ensembling and Blending
  • Venue: Ideation Room at the Manoa Innovation Center

Week 5a, Monday April 30, 2:00 PM: End Competition

  • Description: End competition on Monday at 2:00 PM HST (Tuesday 00:00 UTC). No new submissions will be allowed. Participants will be notified of the competition’s end. The top 3 teams will be asked to submit their code and prepare their presentations.
  • Venue: None

Week 5b, Friday May 4, 6:00 PM: Winner’s Presentations

  • Description: Finalize the competition results. Judges will explain their choice for Best Kernel. Winners will present their solutions (3rd → 2nd → 1st).
  • Venue: Presentation Room at Manoa Innovation Center

We may host a follow-up meetup for any participant interested in presenting their solution or kernel.

Sponsors

Special thanks to the HTDC who is generously providing the venue for this event.

HTDC is a dynamic state agency responsible for diversifying Hawaii’s economy developing a flourishing technology industry that provides quality, high-paying jobs for Hawaii residents. HTDC aims to accelerate the growth of Hawaii’s technology industry by providing capital, building infrastructure and developing talent to foster innovation. HTDC is attached to the Department of Business, Economic Development and Tourism (DBEDT).