Spatial uncertainty propagation in ICT data analysis

HAL Id: hal-02605894

https://hal.inrae.fr/hal-02605894

Submitted on 6 Jul 2020

HAL is a multi-disciplinary open access

archive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Spatial uncertainty propagation in ICT data analysis

Maxime Lenormand, Thomas Louail, Marc Barthélemy, J.J. Ramasco

To cite this version:

Maxime Lenormand, Thomas Louail, Marc Barthélemy, J.J. Ramasco. Spatial uncertainty propaga-tion in ICT data analysis. Urban Net 2016, Sep 2016, Amsterdam, Netherlands. pp.18. �hal-02605894�

Maxime Lenormand | Irstea, France

21 September 2016

UrbanNet 2016 | Amsterdam

Spatial uncertainty propagation in

ICT data analysis

Thomas Louail | IFISC, Spain Marc Barthelemy | CEA, France

Motivation

"Reality"

Data

(Spatial) Uncertainty Propagation

Example: population size of Netherlands. We may estimate

it as 16,800,000 milion. Perhaps in reality it is

16,967,234; hence error of -1%

(Spatial) Uncertainty Propagation

Example: population size of Netherlands. We may estimate

it as 16,800,000 milion. Perhaps in reality it is

16,967,234; hence error of -1%

Error is usually not know because reality is not know

Crosschecking information

Crosschecking mobility information

Lenormand et al. (2014) Cross-checking different sources of mobility information.

PlosOne, 9(8):e105407. ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 10−4 ₁₀−3 ₁₀−2 ₁₀−1 ₁₀0 10−4 10−3 10−2 10−1 100

Commuters (Mobile Phone)

C om m ut er s (T w itt er ) ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● 10−4 ₁₀−3 ₁₀−2 ₁₀−1 10−4 10−3 10−2 10−1

Commuters (Mobile Phone)

C om m ut er s (C en su s) ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 10−4 ₁₀−3 ₁₀−2 ₁₀−1 10−4 10−3 10−2 10−1 Commuters (Twitter) C om m ut er s (C en su s) 10−7₁₀−6₁₀−5₁₀−4₁₀−3₁₀−2₁₀−1 10−7 10−6 10−5 10−4 10−3 10−2 10−1 Commuters (Census) C om m ut er s (B B V A )

(Spatial) Uncertainty Propagation

Example: population size of Netherlands. We may estimate

it as 16,800,000 milion. Perhaps in reality it is

16,967,234; hence error of -1%

Error is usually not know because reality is not know

Crosschecking information

Uncertainty propagation analysis

Lenormand et al. (2016) Is spatial information in ICT data reliable?

300,000 mobile phone users' trajectories x 25 two-week periods

Inferring land use and identifying home-work locations

from mobile phone activity

Time (hours)

6 12 18 0 6 12 18 0 6 12 18 0 6 12 18 0 6 12 18 0 6 12 18 0 6 12 18

Lenormand et al. (2015) Comparing and modeling land use organization in cities.

Royal Society Open Science 2, 15052015.

Land use detection

Extraction of 50 independent samples based on

150,000 users activity during on week

Land use detection

0.20 0.25 0.30 0.35 0.40 0.45 0.20 0.25 0.30 0.35 0.40 0.45 0.20 0.25 0.30 0.35 0.40 0.45Residential 0.15 0.20 0.25 0.30 0.35 0.40 0.15 0.20 0.25 0.30 0.35 0.40 0.15 0.20 0.25 0.30 0.35 0.40 Fr action of mobile phone user Business 0.30 0.35 0.40 0.45 0.50 0.55 0.30 0.35 0.40 0.45 0.50 0.55 0.30 0.35 0.40 0.45 0.50 0.55 6 12180 61218 0 612180 6 12180 612180 61218 0 61218 Nightlife Time of day

Land use detection

0 10 20 30 40 50 60

Residential Business Nightlife Other

P

ercentage of total surf

ace

Land use type (a) 60 70 80 90 100 0.00 0.05 0.10 0.15

Shared surface area (%)

PDF (b) Residential Business Nightlife Total

Land use detection

50 100 150 200 250 300 65 70 75 80 85 90 95 Number of users (x 103₎ Shared surf ace area (%) Voronoi Grid 1km Grid 2km Grid 3km 1 2 3 4 5 6 7 8 9 10 11 12 12 11 10 9 8 7 6 5 4 3 2 1

Four weeks period

Four w eeks per iod −15 −10 −5 0 5 S − S

Song et al. (2010) Limits of predictability in human mobility. Science 327, 1018-1021. 52% 5% 6% 27%

Home

Most frequented location between 7pm and 7am

Work

Most frequented location between 8am and 5pm on weekdays

Origin-Destination Matrix

Tij: number of individuals living in

cell i and working in cell j

Most frequented locations

Most frequented locations

50 100 150 200 250 300 0.65 0.70 0.75 0.80 0.85 0.90 0.95 Number of users (x 103₎ λc (a) 50 100 150 200 250 300 0.4 0.6 0.8 Number of users (x 103₎ λc * (b) 50 100 150 200 250 300 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Number of users (x 103₎ λl (c) 50 100 150 200 250 300 0.994 0.995 0.996 0.997 0.998 Number of users (x 103₎ λl (d) Voronoi Grid 1km Grid 2km Grid 3km

Most frequented locations

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

Four weeks period

Four w eeks per iod −0.06 −0.04 −0.02 0.00 0.02 λc− λc

Robustness of home-work locations identified for each user

Most frequented locations

5 10 15 20 25 74 76 78 80 82 84 86 88 Nh Accur acy Home Work

Good agreement between land uses identified from 100,000 users activity signals, with an average of 75% of shared surface area.

Uncertainty on the journey-to-work commuting network is highly dependent of the spatial resolution.

More studies in this spirit need to be done to assess the biases and uncertainty associated with this new data sources.

Take home messages...

Lenormand et al. (2016) Is spatial information in ICT data reliable?

Self-promotion