Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
Waseda University Data Science Competition 2019
Objectives/目的
• To build your data science skills by making the most accurate prediction for Japan’s 2019 Upper House Election (scheduled in July).
• 2019年の参議院選挙の正確な予測を行うことを通じて、データサイエンスの手法を学ぶ
Timeline• 17 days before the Election: Election officially
called
• June 26th: Deadline for team entry
• One day before the Election (July 13 or 20): Deadline for file submission
• July 14 or 21: Election day
• July 27: Presentations, award ceremony, reception
スケジュール投票日の17日前:選挙の公示
2019年6月26日(水):チームエントリー締め切り
7月13日あるいは20日:各種ファイル提出期限
7月14日あるいは21日:投票日(暫定)
7月27日(土)午後:発表会、授賞式、懇親会
Data
• Any publicly available data can be used; No data will be provided by the organizer.
• Should not violate any copyright or image rights; web scraping should be “ethical” and legal.
• Contact the organizer when in doubt.
データ• あらゆる公開データ(早稲田の学生であれば誰にでも無償でアクセスが可能なもの)が利用可能;主催者側からのデータ提供は政党と候補者リストのみ
• 著作権や肖像権の侵害、複製禁止データの公開などは不可;スクレイピングは法的に問題がなく、相手側に負担をかけないなどethicalな方法である必要
• 判断がつかない際は必ず主催者に問い合わせること
Submission Files
1. Two CSV files (PR and District files) with predictions,
2. Script/code used to create your model(s),
3. Slides (PowerPoint or PDF), in English or Japanese, describing data and methods. Up to 8 pages. They will be used for presentations on July 27.
提出ファイル
1. 予測結果を入力した2つのCSVファイル
2. 予測に用いた分析コード・スクリプト
3. データと分析方法を説明したスライド(PPTもしくはPDF、8枚以内、7/27の発表会で使用)
File Submission• Online file submission at dswaseda.com
• Deadline:
• Election on July 14th —> 23:59 JST July 13th
• Election on July 21st —> 23:59 JST July 20th
• Multiple submissions allowed before the deadline; the latest submission used for evaluation.
• The system will not accept late submission
ファイルの提出方法• イベントHP上にアップロード;それ以外の方法は不可
• 提出期限(厳守):
• 7/14 が投票日の場合:7/13 日本時間23:59
• 7/21が投票日の場合:7/20 日本時間23:59
• 提出期限以前であれば、複数回提出が可能;一番最新の提出ファイルを審査に用いる
• 提出期限を過ぎてからのアップロードはシステム上不可
Important Note
• You are NOT allowed to publicly disclose your election forecasts until the election is over, as it will violate the Public Offices Election Law (Article 138, Paragraph 3).
• The violation of the Law can result in up to 300,000 yen of fines or up to two years in imprisonment (Article 242, Paragraph 2).
重要な注意事項
• 公職選挙法に抵触するため、選挙が終わるまで選挙予測を公開することは一切不可(公職選挙法138条3)。
• 違反した場合:二年以下の禁錮又は三十万円以下の罰金(公職選挙法242条2)
Evaluation
• The accuracy of a team’s submission is evaluated on how close they are to the actual election results – i.e. the number of seats correctly predicted.• In addition to accuracy of prediction, teams are also
evaluated by a panel of judges for the quality of their model (including factors such as innovation and general applicability) and their presentation.
45 Districts (43 prefectures + 2 merged ones) Proportional Representation (national)
74 seats 50 seats
124 seats
45 Districts (43 prefectures + 2 merged ones)
74 seats
candidate_J candidate_E outcome
��� Ishida Ichiro 1
�� Yamada Taro 0
�� �� Sasaki Aiko 1
���� Jiro Sato 1
��� Shigeru Kato 0
Hokkaido (3 seats)
45 Districts (43 prefectures + 2 merged ones)
74 seats
candidate_J candidate_E outcome result_d
��� Ishida Ichiro 1 0
�� Yamada Taro 0 1
�� �� Sasaki Aiko 1 1
���� Jiro Sato 1 1
��� Shigeru Kato 0 0
Hokkaido (3 seats)
45 Districts (43 prefectures + 2 merged ones)
74 seats
candidate_J candidate_E outcome result_d point
��� Ishida Ichiro 1 0 0
�� Yamada Taro 0 1 0
�� �� Sasaki Aiko 1 1 1
���� Jiro Sato 1 1 1
��� Shigeru Kato 0 0 0
Hokkaido (3 seats)
2 points
45 Districts (43 prefectures + 2 merged ones)
74 seats
candidate_J candidate_E outcome result_d point
���� Ishida Ichiro 1 0 0
���� Yamada Taro 0 1 0
���� Sasaki Aiko 1 1 1
��� Jiro Sato 1 1 1
��� Shigeru Kato 0 0 0
Hokkaido (3 seats)
2 points
� ���� ���� � ���� ���� ������� �������� �����
���� � � �� ������ � � �
� �� � ��������� � � �
��� ���� � � ��� � � �
Aomori (1 seat)
1 point
We repeat this for 45 districts and the total numberof correctly predicted candidates will be your pointsN.B. You may only predict 74 candidates to win
Proportional Representation (national)
50 seats
party_E vote_share seat_simulated result_pr seats
Liberal Democratic Party 36 34
Constitutional Democratic Party of Japan 20 22
Democratic Party for the People 8 8
Komeito 12 12
Japan Innovation Party 12 12
Japanese Communist Party 10 10
Social Democratic Party 2 2
Total 100% 50 seats 100% 50 seats
Proportional Representation (national)
50 seats
party_E vote_share seat_simulated result_pr seats
Liberal Democratic Party 36 18 34 17
Constitutional Democratic Party of Japan 20 10 22 11
Democratic Party for the People 8 4 8 4
Komeito 12 6 12 6
Japan Innovation Party 12 6 12 6
Japanese Communist Party 10 5 10 5
Social Democratic Party 2 1 2 1
Total 100% 50 seats 100% 50 seats
We calculate the number of seats using the D’Hondt method
Proportional Representation (national)
50 seats
party_E vote_share seat_simulated result_pr seats
Liberal Democratic Party 36 18 34 17
Constitutional Democratic Party of Japan 20 10 22 11
Democratic Party for the People 8 4 8 4
Komeito 12 6 12 6
Japan Innovation Party 12 6 12 6
Japanese Communist Party 10 5 10 5
Social Democratic Party 2 1 2 1
Total 100% 50 seats 100% 50 seats
We calculate the number of seats using the D’Hondt method
seat_simulated
18
10
4
6
6
5
1
50 seats
seats
17
11
4
6
6
5
1
50 seats
Absolute difference
1
1
0
0
0
0
0
Proportional Representation (national)
50 seats
party_E vote_share seat_simulated result_pr seats
Liberal Democratic Party 36 18 34 17
Constitutional Democratic Party of Japan 20 10 22 11
Democratic Party for the People 8 4 8 4
Komeito 12 6 12 6
Japan Innovation Party 12 6 12 6
Japanese Communist Party 10 5 10 5
Social Democratic Party 2 1 2 1
Total 100% 50 seats 100% 50 seats
We calculate the number of seats using the D’Hondt method
seat_simulated
18
10
4
6
6
5
1
50 seats
seats
17
11
4
6
6
5
1
50 seats
Absolute difference
1
1
0
0
0
0
0
2
The sum of absolute differences of each party will be divided by 2
2/2=1 (PENALTY POINT)!"#$
% &"2
Proportional Representation (national)
50 seats
party_E vote_share seat_simulated result_pr seats
Liberal Democratic Party 36 18 34 17
Constitutional Democratic Party of Japan 20 10 22 11
Democratic Party for the People 8 4 8 4
Komeito 12 6 12 6
Japan Innovation Party 12 6 12 6
Japanese Communist Party 10 5 10 5
Social Democratic Party 2 1 2 1
Total 100% 50 seats 100% 50 seats
We calculate the number of seats using the D’Hondt method
seat_simulated
18
10
4
6
6
5
1
50 seats
seats
17
11
4
6
6
5
1
50 seats
Absolute difference
1
1
0
0
0
0
0
2
The sum of absolute differences of each party will be divided by 2
2/2=1 (PENALTY POINT)
This PENALTY POINT will be subtracted from the maximum points of 50
50-1=49
!"#$
% &"2
Presentations on 7/27
• All teams make a poster presentation; top teams (announced on July 27) make an oral presentation with slides (8 minutes/team; strictly enforced).
• Evaluated by judges (event organizers and data scientists of sponsor companies); posters also evaluated by members of the audience.
発表会(7/27)について
• すべてのチームがポスター発表をする必要。トップチーム(当日発表)は口頭によるプレゼンテーション(各チーム8分間、厳守)
• プレゼンテーションは審査員(企画担当教員とスポンサー企業のデータサイエンティスト)が審査。ポスター発表については一般聴衆による投票も実施。
Prizes
• Grand Prize: 100,000 yen
• SPSE Prize: 50,000 yen (reserved for a team consisting of SPSE students)
Note:Team(s) that receive the Grand and SPSE Prizes are not eligible for other prizes.
賞• 最優秀賞:賞金10万円
• 政治經濟學會賞:賞金5万円(政治経済学術院の学生で構成された最優秀チームに授与)
Sponsor Prizes (Tentative)• PR Prize/比例部門
• District Prize/選挙区部門
• Presentation Prize/プレゼンテーション賞
• Poster Prize/ポスター賞
• Prize for undergraduates/high school students/学部生・高校生部門
• Diversity Prize/ダイバーシティ賞
Sponsor Prizes
Team Entry (May 17-)• Required member information:
• Name; Waseda ID number; Department/School; Year/Position; Waseda email address
• Required team information:
• Team name
• Team leader (has to be one of the team members)
• Seminar name (if all members are from the same seminar)
• Diversity statement (if applicable)
チームエントリー (5月17日開始)• メンバー情報:氏名、学籍番号、学部など所属、学年、早稲田メールアドレス(高校生以外)
• チーム情報:
• チーム名
• チームリーダー名(メンバー中より一名選出)
• ゼミ名(全員が同じゼミに所属している場合)
• 多様性についての説明(該当する場合のみ)
Terms and Conditions• All members must agree:
• to follow the rules of the competition
• not to violate copyrights or image rights
• not to publish predictions prior to the election day
Note: this is just a summary; carefully read the full text when you sign up.
誓約書• チームエントリーの際にはメンバー全員が以下の項目に同意する必要(抜粋):
• ルールを厳守すること
• 肖像権や著作権などを侵害しないこと
• 予測は選挙が終わるまでいかなる形でも公開しないこと
Diversity Statement
A special prize may be awarded to a strongly-performing team which includes a mixture of members of different genders, nationalities, ages, or academic backgrounds, or which includes members of minority groups.
多様性について
• 多様性の定義:異なるジェンダー、国籍、年齢、学問領域で構成される、マイノリティグループに属するメンバーを含むなど
• 多様性を持つチームが対象となる特別賞:優秀な成績を残したチームに授与される可能性