Towards efficient prospective detection of multiple spatio-temporal clusters Bráulio Veloso, Andréa Iabrudi and Thais Correa. Universidade Federal de Ouro Preto – UFOP November, 2013, Campos do Jordão, SP – Brazil XIV Brazilian Symposium on GeoInformatics
141
Embed
Towards efficient prospective detection of multiple spatio -temporal clusters
XIV Brazilian Symposium on GeoInformatics. Towards efficient prospective detection of multiple spatio -temporal clusters. Bráulio Veloso , Andréa Iabrudi and Thais Correa. Universidade Federal de Ouro Preto – UFOP November, 2013, Campos do Jordão , SP – Brazil. Content. Introduction - PowerPoint PPT Presentation
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Towards efficient prospective detection of multiple
spatio-temporal clusters
Bráulio Veloso, Andréa Iabrudi and Thais Correa.Universidade Federal de Ouro Preto – UFOPNovember, 2013, Campos do Jordão, SP – Brazil
XIV Brazilian Symposium on GeoInformatics
Content
• Introduction• Method– STCD– Problem– STCD-Sim
• Metrics• Simulated Datasets• Results• Final Considerations
Introduction
• Technique to efficiently detect multiple emergent clusters in a space-time point process
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Introduction
• Technique to efficiently detect multiple emergent clusters in a space-time point process– Surveillance Systems;• On-line;• Prospective;
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Introduction
• Technique to efficiently detect multiple emergent clusters in a space-time point process– Surveillance Systems;– Applications:
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Introduction
• Technique to efficiently detect multiple emergent clusters in a space-time point process– Surveillance Systems;– Applications:• Epidemic surveillance;
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Introduction
• Technique to efficiently detect multiple emergent clusters in a space-time point process– Surveillance Systems;– Applications:• Epidemic surveillance;• Criminology behavior;
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Introduction
• Technique to efficiently detect multiple emergent clusters in a space-time point process– Surveillance Systems;– Applications:• Epidemic surveillance;• Criminology behavior;• Traffic control;
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Introduction
• Technique to efficiently detect multiple emergent clusters in a space-time point process– Surveillance Systems;– Applications:• Epidemic surveillance;• Criminology behavior;• Traffic control;• Social networks behavior;
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Introduction
• Technique to efficiently detect multiple emergent clusters in a space-time point process– Spatio-temporal data are more available;
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Introduction
• Technique to efficiently detect multiple emergent clusters in a space-time point process– Spatio-temporal data are more available;– Process with more then one cluster;
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Introduction
• Technique to efficiently detect multiple emergent clusters in a space-time point process– Spatio-temporal data are more available;– Process with more then one cluster;– Need of computationally efficient approaches.
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Introduction
• STCD– Point Process;
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Introduction
• STCD– Point Process;– Earlier identification;
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Introduction
• STCD– Point Process;– Earlier identification;– Fast Execution;
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Introduction
• STCD– Point Process;– Earlier identification;– Fast Execution;– Efficient detection;
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Introduction
• STCD– Point Process;– Earlier identification;– Fast Execution;– Efficient detection;
– But identifies only one cluster.
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
The Space-Time Cluster Detection
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
• Renato Assunção and Thais Correa. Surveillance to detect emerging space-time clusters. Computational Statistics and Data Analysis, 53(8):2817-2830, 2009.
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
• Surveillance Systems
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
• Surveillance Systems– Process: Under Control vs. Out of Control;– System: try to detected earlier a change in the
process
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
• Surveillance Systems;• Spatio-Temporal Events – Tuple: (id, x, y, t);– Order by time.
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
• Surveillance Systems;• Spatio-Temporal Events;• Alarm– Evidence that the process changed from in control
to out of control.
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
• Surveillance Systems;• Spatio-Temporal Events;• Alarm;• Space-Time Cluster– Cylindrical shape• Circular base in space• Temporal Height
Space
Time
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection– Ck,n : candidate cylinder to be a cluster, beginning
(centered) in event k and ending in the last event
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection– Ck,n : candidate cylinder to be a cluster, beginning
(centered) in event k and ending in the last event;
– Lk : likelihood of the space-time Poisson process when there is a cluster Ck,n;
– L ∞ : likelihood of the space-time Poisson process when there is no cluster.
• a
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
LLk
STCD – Space Time Cluster Detection• Cumulative Sum Statistic
n
knk
n
k
kn L
L=R1
,1
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Cumulative Sum Statistic
n
knk
n
k
kn L
L=R1
,1
)(exp1 ,)(
,,
nkCN
nk Cεε+ nk
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
• Each parcel k is related to a candidate cluster.
)(exp1 ,1
)( ,nk
n
=k
CNn Cεε+=R nk
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
• Each parcel k is related to a candidate cluster.
– ε: increase in the intensity inside the cluster Ck,n;
)(exp1 ,1
)( ,nk
n
=k
CNn Cεε+=R nk
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
• Each parcel k is related to a candidate cluster.
– ε: increase in the intensity inside the cluster Ck,n;
– N(Ck,n): number of events inside Ck,n;
)(exp1 ,1
)( ,nk
n
=k
CNn Cεε+=R nk
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
• Each parcel k is related to a candidate cluster.
– ε: increase in the intensity inside the cluster Ck,n;
– N(Ck,n): number of events inside Ck,n;
– μ(Ck,n): expected number of events inside Ck,n.• non parametric estimate for μ(Ck,n).
)(exp1 ,1
)( ,nk
n
=k
CNn Cεε+=R nk
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
• Each parcel k is related to a candidate cluster.
– ε: increase in the intensity inside the cluster Ck,n;
– N(Ck,n): number of events inside Ck,n;
– μ(Ck,n): expected number of events inside Ck,n.• non parametric estimate for μ(Ck,n).
)(exp1 ,1
)( ,nk
n
=k
CNn Cεε+=R nk
n
ttANttkBNC nkn
nk
],(.],(),()(ˆ 0
,
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
• Alarm or not?
– A and‘
ARn
n
knkn =R
1,
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
Space
Timetactual
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
Space
Timetactual
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
Space
Timetactual
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
Space
Timetactual
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
Space
Timetactual
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
Space
Timetactual
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
Space
Timetactual
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
Space
Timetactual
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
Space
Timetactual
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
tactual
Space
Time
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
Space
Timetactual
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
Space
Timetactual
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
Space
Timetactual
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
Space
Timetactual
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
Space
Timetactual
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Parameters: – Spatial Radius (ρ)• Used in the definition of spatial neighborhood for each
event
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Parameters: – Spatial Radius (ρ)• ρ
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Parameters: – Spatial Radius (ρ)• ρ ↑
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Parameters: – Spatial Radius (ρ)• ρ ↑↑
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Parameters: – Spatial Radius (ρ)– Increase in the intensity inside the cluster (ε);
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Parameters: – Spatial Radius (ρ);– Increase in the intensity inside the cluster (ε)• ε
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Parameters: – Spatial Radius (ρ);– Increase in the intensity inside the cluster (ε)• ε ↑
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Parameters: – Spatial Radius (ρ);– Increase in the intensity inside the cluster (ε)• ε ↑↑
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Parameters: – Spatial Radius (ρ);– Increase in the intensity inside the cluster (ε);– Threshold (A)
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Parameters: – Spatial Radius (ρ);– Increase in the intensity inside the cluster (ε);– Threshold (A)• A↓• Faster Detection
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Parameters: – Spatial Radius (ρ);– Increase in the intensity inside the cluster (ε);– Threshold (A)• A↓↓• Increase the number of false alarms
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Parameters: – Spatial Radius (ρ);– Increase in the intensity inside the cluster (ε);– Threshold (A)• A ↑• Slower Detection
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Parameters: – Spatial Radius (ρ);– Increase in the intensity inside the cluster (ε);– Threshold (A)• A ↑↑• No Detection
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Parameters: – Spatial Radius (ρ);– Increase in the intensity inside the cluster (ε);– Threshold (A)• How much events the user wants to wait before a false alarm.
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Operations at time i:
i
Space
Time
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Operations at time i:
i
Space
Time
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Operations at time i:
N(C1,12)
i
Space
Time
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Operations at time i:
μ(C1,12)
i
Space
Time
N(C1,12)
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Operations at time i:
i
Space
Time
μ(C2,12)
N(C2,12)
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Operations at time i:
i
Space
Time
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Operations at time i:
i
Space
Time
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Operations at time i:
i
Space
Time
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Operations at time i:
i
Space
Time
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Operations at time i:
i
Space
Time
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Operations at time i:
i
Space
Time
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Operations at time i:
i
Space
Time
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Operations at time i:
i
Space
Time
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Operations at time i:
i
Space
Time
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Operations at time i:
i
Space
Time
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Operations at time i:
i
Space
Time
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
• Alarm!
– A
ARn
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection
• Alarm!
– A
• Identifying the cluster– a
– Cylinder:
ARn
nknknk 1,max ,*,
nkC *,
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD – Space Time Cluster Detection• Operations at time i:
Space
Time
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Problem• Is there more than one cluster?
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Problem• Is there more than one cluster?
Space
Time
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Problem• Is there more than one cluster?
Space
Time
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Space
Time
Problem• How to identify these two clusters
simultaneously?
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
STCD-Sim
Our extension
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
• Clusters:– ε = 1.0, 3.0 e 10.0;– ρ = 0.5, 1.0, 1.5 e 2.0;– Δt = [5, 10], [7, 10] e [8, 10];• The process begin under control in time 0 and one or
two clusters start at time 5, for example.
• Running the STCD:– Input parameters (equal to the true values);– A = n (total number of events).
Simulated databases| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Simulated databases
Results
Percentage of AlarmsDelays
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Results – Alarms unique cluster
• No Alarm
• Incorrect Alarm
• Correct Alarm
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Results – Alarms unique cluster
• No Alarm
• Incorrect Alarm
• Correct Alarm
General Mean:1.37%; 3.39%; 95.24%
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Results – Alarms two clusters
• No Alarm
• Incorrect Alarm
• Incomplete Alarm
• Complete Alarm
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Results – Alarms two clusters
• No Alarm
• Incorrect Alarm
• Incomplete Alarm
• Complete Alarm
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
General Mean:• 1.00%• 1.69%• 63.68% • 33.63%
Results – Alarms two clusters
• No Alarm
• Incorrect Alarm
• Incomplete Alarm
• Complete Alarm
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
General Mean:• 1.00%• 1.69%• 63.68% • 33.63%
Our extension reached a complete Alarm in 88.2% of cases in database
Results – Delay
• Delay 1C.
• 2C. Delay 1st
• 2C. Delay C1
• 2C. Delay C2
• 2C. Delay Duplo
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Results – Delay
• Delay 1C.
• 2C. Delay 1st
• 2C. Delay C1
• 2C. Delay C2
• 2C. Delay Duplo
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Final Considerations
• Our extension for multiple cluster– Percentage of detection for both clusters around
88%;
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Final Considerations
• Our extension for multiple cluster– Percentage of detection for both clusters around
88%;– Delay for two clusters slightly larger than delay for
one cluster;
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Final Considerations
• Our extension for multiple cluster– Percentage of detection for both clusters around
88%;– Delay for two clusters slightly larger than delay for
one cluster;– Promising method.
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Final Considerations
• Future works– Evaluate the impact of changing ρ and ε;
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Final Considerations
• Future works– Evaluate the impact of changing ρ and ε;– Apply to a real database and benchmark;
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Final Considerations
• Future works– Evaluate the impact of changing ρ and ε;– Apply to a real database and benchmark;– Compare with others approaches;
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
Final Considerations
• Future works– Evaluate the impact of changing ρ and ε;– Apply to a real database and benchmark;– Compare with others approaches;– Remove the restriction of the cylindrical shape for
the cluster.
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |
References[1] Renato Assunção and Thais Correa. Surveillance to detect emerging space-time clusters. Computational Statistics and Data Analysis, 53(8):2817-2830, June 2009.
[2] B. Veloso, A. Iabrudi and T. Correa. Localização em tempo real de acontecimentos através de vigilância espaço-temporal de microblogs. In IX Encontro Nacional de Inteligência Artificial, 12 pages, Curitiba - PR, Brazil, October 2012.
[3] C. Sonesson and D. Bock. A review and discussion of prospective statistical surveillance in public health. Journal of the Royal Statistical Society: Series A (Statistics in Society), 166(1):5–21 , 2003.
[4] M. Höhle. surveillance: An R package for the monitoring of infectious diseases. Computational Statistics, 22:571–582, 2007.
| Introduction | STCD | Problem | STCD-Sim | Metrics | Datasets | Results | Final Considerations |