A A A+ A++

AI Challenge – Weather Forecasting Competition

Echoing the theme of 'Master Code 2018' - Be a Smart eCitizen for 2030+, Hong Kong Education City (HKEdCity), Microsoft, Hong Kong Observatory Headquarter and Hong Kong Meteorological Society jointly organise the 'AI Challenge – Weather Forecasting Competition' for Secondary school students. Participants will make use of machine learning/deep learning in AI and the previous 10 years’ meteorological data provided by the Hong Kong Observatory to forecast temperature at Zero Carbon Building (ZCB) in Kowloon Bay in a specific period.

 

Update:

2018/12/7

Results of the competition have been announced, please view it here.

 

(2018/11/29 18:40 PM)

ZCB data update has resumed at 5 PM. The organisers have also retrieved part of the previously missing data, which can be downloaded at the data link: aka.ms/hkochallenge. However, part of the data is not able to be retrieved. Therefore, some of the prediction hours will be voided. Please find the details below:

Retrieved data: 20181128 2001 to 20181129 0501

Data cannot be retrieved: 20181129 0601 to 20181129 1601

No score will be calculated for 20181129 1201-1601 (5 data points). 0 penalty points will be awarded to all submitted forecasts in this period while the rest of the submitted forecasts remain valid.

 

 

(2018/11/29 12:50 PM)

The organisers are aware of the situation for missing ZCB data since last night (28 Nov 2018). A preliminary check showed that the problem was probably related to the sensor at ZCB. There is a chance of missing data for some hours.  An investigation is being conducted and updates will be announced before 5 PM today

Meanwhile, all submitted forecasts for today remain valid.  The organisers will assess the effect of the problem and decide if certain prediction hours would be voided.

Please stay tuned for our updates, sorry for the inconvenience caused.

 

(2018/11/26 12:00 Noon)

Please note that the Longitude of ZCB stated on the HKOData-Readme.docx (from the Data provided by Hong Kong Observatory) is incorrect, the correct longitude is 114 12 30.

Please use the correct longitude in the upcoming temperature forecast.

 

Submission of Temperature Forecast (starting from 12:00 noon, 25 November)

 

Submission of introductory PowerPoint and video (starting from 12:00 noon, 25 November)

 

Objectives

  • To empower teachers and secondary school students to make use of machine learning/deep learning in AI
  • To let students experience how to solve real-life problems using AI
  • To enhance students’ awareness of climate change and involve them to shoulder the responsibility for driving sustainable development of our globe

Target

  • Secondary School Students

Rules & Regulations

  • Each school can form two participating teams maximum, and each team should consist of two to five students. Each student can only participate in one team.
  • From 26 to 30 November 2018, the participating teams need to submit the hourly temperature forecast from 12 noon to 11 pm at Zero Carbon Building (ZCB) for each day (i.e. 12 data per day, 60 data altogether). The submitted forecast temperature at ZCB is in degree Celsius, good to 0.1°C
  • Participating teams need to submit:
    1. The hourly temperature forecast from 12 noon to 11 pm at Zero Carbon Building (ZCB) for each day (i.e. 12 data per day, 60 data altogether)
      • Submission Time: Daily forecast can be submitted from 12:00 noon on the day before to 10:00 am on the forecast day. (E.g. submission period of the forecast of 26 Nov:  25 Nov 12:00 noon to  26 Nov 10:00 am)
      • The submitted forecast temperature at ZCB is in degree Celsius, good to 0.1°C
      • Each team can submit once only per day, no changes can be made after the submission.
    2. An introductory PowerPoint (within ten slides) AND an introductory video (within three minutes) ON OR BEFORE 30 November to elaborate how machine learning / deep learning of AI is used in deducing your results for this forecasting competition.
  • Submission of any entry indicates that the participant has read and accepted the Rules & Regulations of the competition.
  • The Organisers own the copyright of the awarded entries.
  • The Organisers reserve the right to amend and/or cancel the activities as well as the Rules & Regulations without prior notice.
  • In case of any dispute, the Organisers reserve the right of final decision.

Registration

The competition registration will start in mid-Oct, please register the briefing session and workshop on 20 Oct first to learn more about the details.

 

Briefing Session and Machine Learning Workshop*

  • Target: Secondary school Teachers and students

  • Venue: Lecture Theatre 2505, AC2, 2/F, Li Dak Sum Yip Yio Chin Academic Building, City University of Hong Kong

  • Date & Time: Oct (Sat)   9:00 am – 11:00 am

  • Content:

    • Introduction of the Challenge
    • Introduction of Machine Learning
    • Azure ML Studio hands-on workshop

 

*Enrolment by school

 

Co-organisers

Microsoft          香港天文台     香港氣象學會     

 

 

Participating teams need to submit:

  1. The hourly temperature forecast from 12 noon to 11 pm at Zero Carbon Building (ZCB) for each day (i.e. 12 data per day, 60 data altogether)
    • Submission Time: Daily forecast can be submitted from 12:00 noon on the day before to 10:00 am on the forecast day. (E.g. submission period of the forecast of 26 Nov:  25 Nov 12:00 noon to  26 Nov 10:00 am)
    • The submitted forecast temperature at ZCB is in degree Celsius, good to 0.1°C
    • Each team can submit once only per day, no changes can be made after the submission.
  2. An introductory PowerPoint (within ten slides) AND an introductory video (within three minutes) ON OR BEFORE 30 November to elaborate how machine learning / deep learning of AI is used in deducing your results for this forecasting competition.

 

Submission of Temperature Forecast (starting from 12:00 noon, 25 November)

Submission of introductory PowerPoint and video (starting from 12:00 noon, 25 November)

 

The judging panel is formed by professionals from coorganisers and follows the below criteria to carry out the judging:

1 Forecast Accuracy 50%
  • The submitted temperature forecast will be compared with HKO instrument's reading at ZCB. The closer to the instrument reading, the less penalty points one will get. For example: temperature forecast at 2 pm was 27.5°C while the instrument reading was 27.2°C. Then, the difference or error would be 0.3°C, which means 0.3 penalty points will be got, and so on.
  • 50 penalty points will be given to all late or nil submission of any single forecast.
  • Maximum penalty for each submission day will also be 50 points.
  • The final score of each team will be obtained after normalisation with all other teams' submissions
2 Application of Machine Learning 20%
  • Demonstrate the application of machine learning algorithms, e.g. regression, cluster analysis, etc.
  • Explain the use of algorithms in the work
3 Application of Data analysis and Computational Thinking 20%
  • Use mathematical methods to estimate average values, maximum and minimum values, etc.
4 Team Collaboration and Communication 10%
  • Clearly show how the team works together by listing the division of labor, etc.

 

 

The below data will be provided to the participants:

Historical data

  • Hourly temperature, humidity, wind speed and direction, rainfall amount from 14 HKO automatic weather stations (AWS) of the past 10 years

  • Hourly temperature data from ZCB for a certain past period

Near real time data (starting from October, updated once a day tentatively)

  • Hourly data from the 14 HKO AWS

  • Hourly data from ZCB

 

Data link: aka.ms/hkochallenge

 

Participants can also use other data from HKO website, MyObservatory mobile app, or any other open data or information from the Internet.

Winners

Awarded schools will be invited to join the award ceremony held in Learning & Teaching Expo at Hong Kong Convention and Exhibition Centre at 3 p.m. on 14 December 2018.

Award School Name Team Members
Champion TWGHs Mrs. Wu York Yu Memorial College 叶融杰、江學明
First Runner-up La Salle College Lee Cheuk Kit, Kwok Long Fung, Mak Sebastian Duncan, Tse Pun Yiu, Yeung Chun Yin
Second Runner-up Po Leung Kuk Tang Yuk Tien College Choi Ho Yeung, Leung Tsan Fung, Chow Man Hei, Chu Yui Lam, Chui Wai Chun Victor
Merit Awards Bishop Hall Jubilee School Lau Wai Ming, Cheung Kin Ho, Ng Cheuk Ting, Wong Tak Ming Damien
Christian Alliance S W Chan Memorial College 梁穎豪、袁竣希、楊天恒、池成傑、鄭明浩
Our Lady of The Rosary College Wong Cheuk Yiu, Poon Hoi Ching Phoebe, Wan Sai Hang, Lau Hoi Ying
SKH Bishop Mok Sau Tseng Secondary School Cheng Pak Yim, Wong Siu Nam, Yip Kwo Fung, Lui Cheuk Hei, Lee Chun Hei 
St. Mark's School Choy Wing Kwan, Ng Hei Wai, Wong Ka Yin, Tsang Yi Tung
St. Paul's College Lee Chi To, Lai Kin Shing, Mang Hao Jian
Stewards Pooi Kei College Cheng King Lam Kris, Lam Wing, Lau Sik Yu, Lam Chun Ho, Tam Hoi Yeung

 

Awards for 'AI Challenge – Weather Forecasting Competition':  

Awards Quantity Prizes
Champion 1 TBC
First Runner-up 1 TBC
Second Runner-up 1 TBC
Merit Awards 7 TBC

 

  • Each participating student who has completed the competition (submitted all 5-day forecast and the files) will be presented a certificate
  • Winners will also be presented relative certificates

Online learning materials

Microsoft - Introduction to Machine Learning (PDF)

Introduction to Azure Machine Learning Studio (Youtube playlist)

 

 

Briefing Session and Machine Learning Workshop*

  • Target: Secondary school Teachers and students

  • Venue: Lecture Theatre 2505, AC2, 2/F, Li Dak Sum Yip Yio Chin Academic Building, City University of Hong Kong

  • Date & Time: Oct (Sat)   9:00 am – 11:00 am

 

Introduction of the Challenge

 

Introduction of Machine Learning

 

 

Azure ML Studio Hand-on

 

Introduction of Hong Kong Observatory's data