mining in the middle of the city: the needs of big data for smart cities
TRANSCRIPT
Dr. Antonio J. Jara – [email protected] HES-SO//Valais Switzerland
Mining in the Middle of the City: The needs of Big Data for Smart Cities
A Real Experience in the SmartSantander Testbed
Antonio J. Jara, Dominique Genoud, Yann Bocchi HES-SO, Switzerland
Palo Alto, USA 19th June 2014
Problem statement
• Smart Cities are presenting new challenges for Big Data. • The emerging amount of data needs to be processed to
make feasible its analysis. • First step, data fusion to avoid noise and apparently
random behaviors. • Second step, correlation in order to see hidden
behaviors. • Next steps more focused on insight, and integration into
business models. • Needs from the market to define the questions that are
expecting to answer for the Smart Cities.
Dr. Antonio J. Jara – [email protected] HES-SO//Valais Switzerland
Big Data / Smart Cities ecosystem
Dr. Antonio J. Jara – [email protected] HES-SO//Valais Switzerland
SmartSantander Testbed
Dr. Antonio J. Jara – [email protected] HES-SO//Valais Switzerland
SmartSantander Testbed (Temperature)
Dr. Antonio J. Jara – [email protected] HES-SO//Valais Switzerland
Data Fusion
• Temperature area totally insolated from the traffic monitoring zones. • Not required fine-grain analysis of temperature, since
not influenced by traffic.
• Traffic sensors needs to be aggregated by highways and lanes.
• Data fusion feasible due to the nature of the problem.
• This simplify and makes feasible the correlation between Temperature and Traffic
Dr. Antonio J. Jara – [email protected] HES-SO//Valais Switzerland
Traffic vs Temperature in April (with data fusion)
Dr. Antonio J. Jara – [email protected] HES-SO//Valais Switzerland
Traffic vs Temperature in July (with data fusion)
Dr. Antonio J. Jara – [email protected] HES-SO//Valais Switzerland
57,4 % Line Correlated
Traffic vs Temperature in December (with data fusion)
Dr. Antonio J. Jara – [email protected] HES-SO//Valais Switzerland
Modelling of Temp / Traffic in April
Dr. Antonio J. Jara – [email protected] HES-SO//Valais Switzerland
Modelling of Temp / Traffic in July
Dr. Antonio J. Jara – [email protected] HES-SO//Valais Switzerland
Modelling of Temp / Traffic in December
Dr. Antonio J. Jara – [email protected] HES-SO//Valais Switzerland
Conclusions • Data Fusion is required for Smart Cities analysis. • Correlation of non-aggregated data is non-feasible. • Data Fusion has demonstrated the similarity among the
temperature and traffic trends. • KNIME offers an intuitive tool to works with Data. • In addition, it offers correlation tools, characterization
tools, and classification tools from Weka and R, and finally with Hadoop.
• Current works focused on human dynamics analysis over the data; Burst vs Poisson.
• An extended / advanced version of this work avaiable under request to [email protected]
Dr. Antonio J. Jara – [email protected] HES-SO//Valais Switzerland