small area Tourism Visible Using Mobile Positioning Data (MPD) for small area (regency/city and venue level) Rifa Rufiadi and Alfatihah Reno (BPS-Statistics Indonesia) rifa@bps.go.id, alfa@bps.go.id
Introduction BPS-Statistics Indonesia never published domestic tourism data at Regency/City level due to expensive and huge work burden, while the data became more important and demanding by policy maker and business. Compared to household survey for domestic tourism, using Mobile Position Data can reduce cost more than 60 %, cover more origins and destinations, get near real time data, get by product data such as commuters and circler travellers. This 2018 year, as pilot we run both ways of collecting domestic tourism data, household survey and Mobile Position Data, while Asian Games event analysis 2018 was done as part of tourism. To cope the limitation of MPD such as no expenditure data, digital survey was conducted using selected sample indicated by Mobile Position Data BPS expects to substitute surveys with timely and more accurate digital data collection in the future
Sample Household Survey Distribution sample 30.000 census block, 50.000 traveller Listing, chose eligible sample, estimate population face to face interview Using MPD & digital survey Distribution sample All census block (more than 800.000) +/- 1.799.342 subscriber All domestic tourism subscribe to MNO, estimate for the non subscriber Ground check
Enumerator Household Survey Minimum : 8.000 x 4 = 32.000 man survey Recruitment Training Questionnaire & manual Office supplies Quarterly, memory laps Resistant from respondent Remote area problem, difficult to be reached MPD & Digital Survey No need enumerator, reduce burden Monthly (timeliness), even daily data Fact, real data (accuracy)
Cost Benefit Household Survey Rp1.300.000,-/ Census block Rp800.000,-/ respondent No learning process MPD & Digital Survey Rp18.750,-/Census Block (1,4%) Rp8.300,-/respondent (1%) Give experiences to BPS s data scientist 1 US $ = Rp 15.000
The Process Prepare the algorithm and scripts Implemented to 66 volunteers Compare Household survey result with MPD Increase data quality (level of disaggregation and accuracy) Event Analysis of Asian Games (foreign and domestic visitors) at the Venue (GBK and JSC)
Algorithm to identify Tourism Trip Generate stay point from raw data Remove usual environment except Home (Work, Second Home (Circular), etc.) Create sequence of trips from stay point with duration > 6 hours Identify each destination visited with duration > 6 hours on each trip Identify the main destination from each trip based on the longest duration of stay Result : - Matrix of main destination with average length of stay - Matrix of all destination with average length of stay 7
WORK Validation/Verification First verification is done to determine where the person is staying. All stay points detected as home locations are asked to be corrected. Further verification is carried out to determine the place of daily activities. All stay points detected as work / school locations are asked to be corrected. 8
Acuracy of Home Detection of the Volunteers Accuracy of home detection is 85,83% (rank 1) Accuracy of home candidate rank 2 : 58,92 % 3 : 58,02 % 4 : 51,50 % 9
PRELIMANRY RESULT: HOME REGENCY POPULATION PROJECTION SUBSCRIBER ONE MNO % TOTAL HOME %
One subscriber Tourism Trip Tourism Trip Work Home
Event Analysis Asian Games
Preliminary Result, Outbound Yogyakarta Province Jan-June 2018 Population Traditional Survey** MPD* ORIGIN Sum % Sum % Sum % Kulon Progo 425.384 11,2 214.611 7,0 1.880.890 10,7 Bantul 1.005.736 26,5 1.337.077 43,4 4.403.341 25,2 Gunung Kidul 735.638 19,4 561.134 18,2 1.331.569 7,6 Sleman 1.205.610 31,7 680.609 22,1 6.520.810 37,3 Yogyakarta 427.099 11,2 283.860 9,2 3.365.504 19,2 TOTAL 3.799.467 100,0 3.077.291 100,0 17.502.114 100,0 * January 2018 ** One shot survey on July 2018
ORIGIN & DESTINATION DOMESTIC TOURISM, YOGYAKARTA PROVINCE JANUARI 2018 ORIGIN DESTINATION GUNUNG KOTA KULON BANTUL KIDUL YOGYAKARTA PROGO SLEMAN Grand Total KOTA YOGYAKARTA 1.867.536 94.799 121.307 2.921.346 5.004.988 BANTUL 459.217 1.161.873 588.903 448.139 2.658.132 SLEMAN 467.851 155.246 1.344.669 632.876 2.600.642 GUNUNG KIDUL 757.171 117.006 8.345 510.147 1.392.669 KULON PROGO 445.436 5.381 65.139 554.440 1.070.396 KLATEN 24.483 194.672 52.846 14.062 127.101 413.164 MAGELANG 23.246 16.257 48.542 190.515 105.753 384.313 KOTA SEMARANG 64.413 34.965 18.198 35.990 146.625 300.191 PURWOREJO 54.518 3.260 23.454 87.652 75.048 243.932 SUKOHARJO 11.784 135.542 22.778 12.185 29.947 212.236 KOTA SURAKARTA 44.651 23.885 16.323 16.409 109.767 211.035 KEBUMEN 41.472 3.788 24.637 5.056 86.228 161.181 BOYOLALI 32.787 7.355 21.814 3.652 82.192 147.800 WONOGIRI 28.556 33.317 17.983 3.029 63.639 146.524 Other Destination 539.437 163.885 430.242 160.909 1.260.438 2.554.911 TOTAL 4.403.341 1.331.569 3.365.504 1.880.890 6.520.810 17.502.114
DOMESTIC TOURISM, ORIGIN YOGYAKARTA PROVINCE & PROVINCE LEVEL DESTINATION JANUARI 2018,
DOMESTIC TOURISM, ORIGIN YOGYAKARTA PROVINCE & REGENCY LEVEL DESTINATION JANUARI 2018,
Summary Monthly data of Domestic Tourism up to sub province Level (even District level) can be published monthly Reduce respondent and work burden Number of trips will be more accurate (no recalling problems) Household survey can only published up to provincial level (sampling issue) and annually. Number of trips possible under (recalling problems), the survey asked number of trips during six months Immigration data can not capture up to venue level (experience from Asian Games)