Name: MARCO AURÉLIO BRUNORO THOMÉ

Publication date: 13/11/2020
Advisor:

Name Rolesort descending
VINICIUS FERNANDES SOARES MOTA Advisor *

Examining board:

Name Rolesort descending
VINICIUS FERNANDES SOARES MOTA Advisor *

Summary: Data gathered by sensors, cameras, social networks, and applications can contribute on
automatic detection of atypical traffic events like accidents and floodings. Moreover, the
heterogeneous nature of a plethora of data sources has the advantage of information
redundancy, which helps increasing the degree of reliability for detected events. In this
work, it is proposed a solution for real-time anomaly detection and notifications, with an
interface that allows heterogeneous data sources. The approach is based in splitted time
series clustering by periods, generating a pattern that is used as a parameter for event
classification, with an outlier detection technique. To evaluate the proposal, a prototype
was developed, as a proof of concept, with deployment of data interfaces for Vitoria’s Waze
based API and Twitter information. Then, the prototype’s performance was analyzed
after varying some parameters, like the clustering algorithm and outlier detection method,
in order to obtain a good configuration set for a definitive system use. By using real
data collected in the city, the results show that the proposed solution can help managers
and agents in their decision making. Keywords: Urban mobility. Clustering. Time series.
Anomaly detection. Real-time.

Access to document

Acesso à informação
Transparência Pública

© 2013 Universidade Federal do Espírito Santo. Todos os direitos reservados.
Av. Fernando Ferrari, 514 - Goiabeiras, Vitória - ES | CEP 29075-910