Detecting Sudden Rises in Traffic Demand Based on Route Search Request Log

Size: px

Start display at page:

Download "Detecting Sudden Rises in Traffic Demand Based on Route Search Request Log"

Ruth Maxwell
5 years ago
Views:

1 CIU 2015 (March 18, 2015) Detecting Sudden Rises in Traffic Demand Based on Route Search Request Log Naoki Nozu, Kohei Ota, Reimi Ishimura NAVITIME JAPAN Co., Ltd.

2 About NAVITIME A Multi-modal Navigation Service Provider on Mobile Phone Route Search Service Automotive Navigation Founded : March 1,

3 Scale of Our Services 4 million Dues-paying Members 26 million Monthly Unique Users As of September 30,

4 Our Business Domain Consumer Services Domestic Apps. Business Solutions Web Media Telematics Inbound App. International Apps. Transport Consulting Business NAVITIME ASP 4

5 Our Data for Traffic Analysis Today s Topic Traffic Demand Big Data Master Data Route Choice Road Traffic Route Search Request Logs Route Choice Logs Road Network Data Travel Record Probe Data at 1 sec. interval GPS Public Transport Timetable, Station Data 5

6 Route Search Request Logs Origin & Destination Time Setting Journey Details Request Time Present Specified Time Future Reflect Future Traffic Demand 6

2 Kanagawa 152.4 mil. 3 Osaka 126.7 mil. 45 Tokushima 0.7 mil. 46 Kochi 0.7 mil. 47 Okinawa 0.3 mil. # Pref. Samples 1 Tokyo 14.3 mil. 2 Kanagawa 9.

7 Data Volume (FY 2013) Public Transport Road Traffic Annual Sample Number less than 3 mil. 3 mil. and more 10 mil. and more 30 mil. and more 100 mil. and more Annual Sample Number less than 1 mil. 1 mil. and more 2 mil. and more 5 mil. and more 10 mil. and more # Pref. Samples 1 Tokyo mil. 2 Kanagawa mil. 3 Osaka mil. 45 Tokushima 0.7 mil. 46 Kochi 0.7 mil. 47 Okinawa 0.3 mil. # Pref. Samples 1 Tokyo 14.3 mil. 2 Kanagawa 9.0 mil. 3 Saitama 8.0 mil. 45 Saga 0.6 mil. 46 Kochi 0.5 mil. 47 Tokushima 0.5 mil. 1.3 bn. Samples a Year 130 mil. Samples a Year Large Amount of Samples Available 7

8 Research Outline Route Search Request Logs Large Amount of Samples Availabe esp. for rail service Reflect Future Traffic Demand Purpose of our Study To Detect Sudden Rises in Traffic Demand by Using Route Search Request Logs 8

9 Heat Map Visualizing the Concentration of Traffic Demand April 13, 2013 Seibu- Kyūjō-mae station 9

10 Heat Map Visualizing the Concentration of Traffic Demand Before the curtain time of Momoiro Clover Z s concert at Seibu Dome, April 13, 2013 Seibu-Kyūjōmae was the 7th most requested station. Central Tokyo 10

The Number of Request Logs Cumulated Number of Route Search

on April 13, 2013 2000 1500 1000 500 0 Before the Opening Time of

before 2 hours before 15 hours before 4 days before Cumulated

day at 4 days before 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

11 The Number of Request Logs Cumulated Number of Route Search Requests Seibu-Kyūjō-mae Sta. on April 13, Before the Opening Time of the Merchandise Booth Before the Curtain Time Real Time 10 min before 2 hours before 15 hours before 4 days before Cumulated Number in an Ordinary Day 8 times more requested than an ordinary day at 4 days before Specified Arrival Time of Route Search Requests Possible Use for the Capacity Adjustment & Retail Marketing 11

Increasing Trend in Advance Searches Advance Search Rate The Growth of Advance Search Gives the Key to the Prediction 100% 80% 60% People run searches for morning trips much earlier than for evening

12 Increasing Trend in Advance Searches Advance Search Rate The Growth of Advance Search Gives the Key to the Prediction 100% 80% 60% People run searches for morning trips much earlier than for evening trips. People run searches by arrival time much earlier than by departure time. 40% [DPT] 6 a.m. 20% [ARR] 6 a.m. [DPT] 8 a.m. 0% 4 days before 15 hours before 2 hours before 10 min before [ARR] 8 p.m. Real Time Time Difference between Request Time and Specified Time People Run Searches in Advance for Morning Trips by Arrival Time 12

13 Detection Method Learning Period : From February 1 To March 17, 2013 (6 Weeks) Target Period for Prediction : From March 18 to April 14, 2013 (4 Weeks) Target Units for Prediction : 1,762,880= 1,574 Stations * 28 Days * 20 Time Zones * 2(Departure or Arrival) 平常日の2 倍以上の検索数が予測されたら突発的移動需要 A sudden rise is defined as a case which retrieval frequencies become at twice an ordinary day? 件過去の傾向を元に検索の伸びを予測 Forecasting the growth of the number of the future route search 平常日の検索数 The basic retrieval frequencies in an ordinary day 4 日前 15 時間前 2 時間前 10 分前検索指定日時検索実施日時 References Ishimura R, Ota K & Tomii N (2013) Detecting Sudden Rise in Demand of Transportation Needs in the Near Future Based on Immediate and Historic Data of Route Search Service (in Japanese). The 47th Research Meeting on Civil Engineering Planning,

14 Detection Result Target Period: From March 18 To April 14, 2013 (4 weeks) The Number of Sudden Rise in Traffic Demand: Total 12,268 (2.2% were detected 4 days before) Cause Event Concert Sporting Event Other Event Holiday Resort Government & Business District School Event Number Detected Stations for Example Seibu-Kyūjō-mae, Suidōbashi, Korakuen, Saitama- Shintoshin, Shin-Yokohama Ukimafunado, Hasune and Keisei-Sakura (for marathon), Urawa-misono Kokusai-tenjijō-seimon, Yokosuka-chūō (for U.S. Fleet Activities) Kudanshita (for cherry-blossom viewing), Takaosanguchi, Tokyo DisneySea Station 36 Nihon-ōdōri, Tochōmae, Kasumigaseki, Nishi-shinjuku 47 Kudanshita, Hiyoshi, Chuo-Daigaku Meisei-Daigaku Time Schedule 15 Wakōshi, Shinjuku-sanchome, Kita-sando, Revision Motomachi-Chūkagai Airport 2 Haneda Airport, Haneda Airport Terminal 1 Unknown 48 Total

15 The World s Fastest? Demand Detection of NEW SHINKANSEN 15

16 Last Weekend Kanazawa Toyama Tokyo 16

17 Station Congestion Alert Station Congestion Prediction Calendar as of March 7 (Japanese) Toyama Station Kanazawa Station 17

18 Kanazawa or Toyama? 18

19 Kanazawa or Toyama? 19

20 The Number of Request Logs 3 月 14 日 3 月 15 日 3 月 16 日 3 月 17 日 3 月 18 日 3 月 19 日 3 月 20 日 3 月 21 日 3 月 22 日 3 月 23 日 3 月 24 日 3 月 25 日 3 月 26 日 3 月 27 日 3 月 28 日 3 月 29 日 3 月 30 日 3 月 31 日 Specified Date (as Arrival) : From March 14 to 31, 2015 Request Date : From February 1 to March 7, 2015 Destination Station : Kanazawa, Toyama Kanazawa Total 23,258 Toyama Total 9, Times as Many as Toyama

21 Origin Distribution to Kanazawa Specified Date (as Arrival): From March 14 to 31, 2015 Request Date: From February 1 to March 7, 2015 Destination Station: Kanazawa Tohoku Shinkansen Area Hokuriku (Nagano) Shinkansen Area Tokyo JR East Widely Draws Customers from their Shinkansen Areas 21

Development Potential The impact to be inscribed

Silk Mill The Daily Number of Searches Origin

Tomioka Silk Mill Sudden Rise after UNESCO s

June 30, 2014 Visiting a Little Further from

群馬サファリパーク Gunma Safari Park Kusatsu 1,000

0 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 2013 2014

22 Development Potential The impact to be inscribed on the UNESCO World Heritage List The Tomioka Silk Mill The Daily Number of Searches Origin Distribution (Public Transport) 2,000 富岡製糸場 Tomioka Silk Mill Sudden Rise after UNESCO s Announcement Target Period: From April 26 To June 30, 2014 Visiting a Little Further from Neighboring Area Minakami Ikaho Nikko 1,500 群馬サファリパーク Gunma Safari Park Kusatsu 1,000 Spin-off Effect on the Safari Park Karuizawa Route Search Request Logs Reflect a Trend Quickly 3~10 ~30 ~100 ~300 ~ ~ 22

Conclusion Sudden Rises in Traffic Demand Have

Route Search Request Logs Detecting Sudden Rises

Transportation Services Retail Marketing

23 Conclusion Sudden Rises in Traffic Demand Have been Difficult to be Detected Can be Detected by Route Search Request Logs Detecting Sudden Rises in Traffic Demand Will Improve Public Transportation Services Retail Marketing Approaches Would Contribute to success 2020 Tokyo Olympics & Paralympics 23

Effects of LCCs Entering Japan s Aviation Market and JAL s Strategy

TOKYO UNIVERSITY Effects of LCCs Entering Japan s Aviation Market and JAL s Strategy Tetsuya Onuki Managing Executive Officer Japan Airlines Co.,Ltd. 08 JAN 2013 Contents 1. Acknowledging Today s Business