INDONESIA’S EXPERIENCE OF USING SIGNALING MOBILE ...

INDONESIA’S EXPERIENCE OF USING SIGNALING MOBILE POSITIONING DATA

FOR OFFICIAL TOURISM STATISTICS

STATISTICS INDONESIA

By

Titi Kanti Lestari, Siim Esko, Erki Saluveer, Sarpono and Rifa Rufiadi

(BPS – Statistics Indonesia and Positium LBS)

STATISTICS INDONESIA

2

INDONESIA AND EUROPE

STATISTICS INDONESIA

3

INDONESIA AND NORTH AMERICA

STATISTICS INDONESIA

BPS rely on the Immigration Record and Border Survey

for Inbound & Outbound Tourism Data

Visits from neighbouring countries only 7% of tourism

Under Coverage :

Not All Border Gates have 24/7 Immigration service

Not All Borders have border gates

Border Survey is too Expensive

4

BACKGROUND

5

Riau islands (SG border) Borneo (MY)

Sintang&Talaud (PH)

Papua (PNG)

East Nusa Tenggara (TL)

Malaysia

Indonesia

Immigration entry points

100 km

Ka

lima

nta

n -

Born

eo

Roamer trip start areahigh low

STATISTICS INDONESIA

7

ALIGNMENT WITH STRATEGY

Big Data- Can be compiled automatically- Real Time- Less Manual Labour

The integration of these new data with traditional data to produce high-quality information that is more detailed, timely, and relevant for many purposes and users, especially to foster and monitor sustainable development.

To measure the mobility of people, including mobility of tourists. The digital footprint left by the users is very sensitive, but also highly valuable, as it provides new possibilities to measure and monitor the spatio-temporal activities of the population.

In line with UN Recommendation “A World that Count”

Data Revolution For Sustainable Development

MPD as one of the Most Promising ICT Data Sources

STATISTICS INDONESIA

CDR – records of active use of mobile phones

Signalling – all signals of mobile phones captured

by the BTS (mobile antenna), even with no call/text

Indonesia uses signaling data due to its precision -

it captures more trips than CDR.

• Signaling data detected more roamers than CDR

(on average 3.47 times).

• So, it overcomes undercoverage issues,

especially in less mobile BTS-dense area.

8

SIGNALLING DATA CAPTURES MORE

STATISTICS INDONESIA

Natuna island

Malaka on the Timor Leste border

Anambas islands

Kupang on the Timor Leste border

Talaud islands

Sangihe islands

Boven Digoel near Papua

1

10

100

1000

1 10 100 1 000 10 000 100 000

Nu

mb

er

of

trip

s fr

om

sig

nal

ling

dat

a (m

ult

iplie

r ra

tio

)

Number of trips from CDR data

Signalling contributes most in hard-to-reach areas

9

SIGNALLING DATA VS CDR

More trips

Even

more

tri

ps

from

sig

naling

da

ta

1. Islands

2. Border to less

developed countries

STATISTICS INDONESIA

10

ROAMING≠TOURISM≠IMMIGRATION

0

2000

4000

6000

8000

10000

12000

14000

16000

18000

20000

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

MPD vs IMMIGRATION, BINTAN, JULY 2017

Immigration MPD

1 456

4384

-

2 000

4 000

6 000

8 000

10 000

12 000

14 000

16 000

18 000

20 000

Total

BINTAN, JULY 1

IMMIGR MPD

618

17624

-

2 000

4 000

6 000

8 000

10 000

12 000

14 000

16 000

18 000

20 000

Total

BINTAN, JULY 23

IMMIGR MPD

?

STATISTICS INDONESIA

Statistical and Non Statistical Noise

Fast fliers

Seamen

Accidental roamers

Other transit

11

SIGNALING DATA ADDS NOISE

MPD

IMMIGRATION

STATISTICS INDONESIA

13

MPD ERROR

0%

600

%1

200% Total improvement in RMSE

0

1

2

3

4

5

6

Error: MPD original Error: LOS1 Error : LOS1 + Seamen excluded Error: LOS1 + Seamen+Fastflyexcluded

RM

SE in

th

ou

san

ds

MPD Error (RMSE) decreases significantly after adjustments

11

x

5x

imp

rove

me

nt

8x

MPD all trips in Bintan vs main port immigration data, rough adjustment with Positium methods

STATISTICS INDONESIA

14

IMMIGRATION DATA VS MPD

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

With and without filters: Tourism arrivals in Bintan

MPD original MPD wo LOS1+seamen+fast flyers Immigration

STATISTICS INDONESIA

Implement MPD Data since October 2016.

Data provided by MNO aggregate table.

Oct-Dec 2016 : MPD was applied to 19 districts,

since January 2017 became 25 districts.

Filtering using a simple 7/20 formula (until Dec 2017), neighbouring countries.

Since January 2018, Filtering MPD has already been using a new formula

Now BPS has an MoU with the biggest MNO in Indonesia, so we can have a

sandbox to process MPD

From 2019, automatic system of processing tourism statistics with all the

necessary algorithms to remove noise

15

CONTINUOUS IMPROVEMENT

STATISTICS INDONESIA

16

MAKING OF

STATISTICS INDONESIA

𝑨𝑻 = 𝑴𝑷𝑫

𝑿𝒓𝒐𝒂𝒎∗ 𝑷 ∗ 𝟏 − 𝑷𝒘 ∗

𝟏

𝟏 − 𝑷𝒏𝒓 − 𝑾𝑪𝑰

FRAMEWORK QUESTIONNAIRE

EXTRAPOLATION FORMULA

STATISTICS INDONESIA

18

EUREKA MOMENT

STATISTICS INDONESIA

19

RESULTS

Number of additional Tourists Jan-July 2018

MPD significantly increases the coverage of inbound tourism in Indonesia.

STATISTICS INDONESIA

20

Indonesia Other countries

7% 30-60%

Before MPD use (2015)

Indonesia Other countries

7% 30% 30-60%

After MPD use (2018)

STATISTICS INDONESIA

21

HANDBOOKS

Methodology Concept and QAF Handbook

Indonesia currently already has draft of Methodology Concept and QAF handbook for the use MPD in Cross Border Inbound Tourism

Standard Framework

GSBPM

Data Collection

Data Processing

System Design

Data Analysis

Methodology Concept

Standard Framework

GSBPM

Quality Gates

Input Quality

QAF Dimension

Output Quality

QAF

THANK YOU

STATISTICS INDONESIA