Transcript Cloud based Big Data Analytics for Smart Future Cities

TRACE ANALYSIS AND
MINING FOR SMART CITIES
By
G. Pan
Zhejiang Univ., Hangzhou, China
G. Qi ; W. Zhang ; S. Li ; Z. Wu ; L. T. Yang
SCOPE
• Analysis and mining of sensed data from dynamic cities is an inevitable step
toward making a city smart
• The trace data can then be input for mining to characterize knowledge
about mobility, people, and the city. Finally, applications can exploit the
knowledge mined to make them smarter in different domains of a smart city.
TRACE DATA
• The main source of Trace data are
1.
Mobile phones (GPS, WiFi, GSM, and Bluetooth)
2.
Vehicles (GPS)
3.
Smart Cards (Bank cards and transportation cards)
4.
Floating sensors (RFID)
CHALLENGES
1.
Mobility
Low-level human activities like step size and orientation is taken into account
but its not enough. The number of locations and frequency of visits should also
be taken into consideration for accurate outcomes.
Transient prediction as simple as predicting the next visited places with
knowledge of historical traces has poor results.
CHALLENGES
2.
HUMAN BEHAVIOR
Low- level human activities, such as walking, sitting, and standing, may be
detected from trace data.
A major challenge for recognizing human behavior is mining high-level
semantics from low-level activities such as where he/she has visited, who
he/she has contacted, what he/she has done, and where he/she will go,
based on his/her trace data.
CHALLENGES
3.
Social Relations
A social relation refers to a relationship among individuals. It includes direct
and indirect inter- actions, ranging from low-level face-to-face interaction to
high-level interactions like reading books or following traditions.
A major challenge for trace-based social analysis is to find potential social
relation in trace data and construct a corresponding social network.
CHALLENGES
4.
CITY DYNAMICS
City Dynamics reflects urban dynamic information in many fields such as
energy consumption, traffic flow, epidemic spread, and urban growth.
A major challenge for those steps is cross-domain inference, which comes
from the heterogeneous nature of trace data.
There are many kinds of trace sources (traces of mobile devices, vehicles,
smart cards, and floating sensors) and localization systems (GPS, Wi-Fi, GSM,
Bluetooth, and payment records)
METHODS FOR TRACE ANALYSIS
AND MINING
• Clustering
Clustering can be used to depict the group of similar traces and patterns for
mining hotspots.
• Classification
Classification is used to predict the individual activity, social event and region
semantics.
• Ranking
Ranking is used to find list of recommending places on the basis of mobility
patterns in descending order.
METHODS FOR TRACE ANALYSIS
AND MINING
• Physical Statistical Model
PSM is used to evaluate human mobility based on patterns from step length,
no of visiting places, visiting frequency and order of visiting patterns.
APPLICATION OF SMART CITIES
• Smart Transportation (Traffic analysis, dynamic dispatch, intelligent
navigation)
• Smart Urban Planning (Guiding, monitoring and evaluating urban plan)
• Smart Public health (Reducing health problems, epidemics and monitoring
behavior of patients)
• Smart Public Security (Detecting misbehavior, social events and tracking
critical people)
CONCLUSION
• Analysis and mining of Trace data is used to detect human behavior and
city dynamics.
• It helps understand human behavior, social events, geographical
importance to make city smarter.
• But still it is still challenging to unify trace data into complex and ever
changing real world.
ANALYSIS OF UNIVERSITY SCORE'S
&
SOFTWARE DEFECT PREDICTION
• By
• Meetkumar Patel
• Srivats Srinivasan
ANALYSIS OF UNIVERSITY SCORE'S
• Purpose – There is a huge amount of data which is unsegregated (19962013). Ability to find a college that is a good fit for the student.
• Scope –
• Integration of the data sets
• Data cleaning
• Web development and queries.
• Result –
• Statistical report’s(State wise/department wise).
• Performance of the University.
SOFTWARE DEFECT PREDICTION
• Purpose – To find what are the major factors that result to a software defect.
• Background –
• NASA Metrics Data Program, have been trying to predict a software's defect using different
algorithm's, and to develop a algorithm which has a 100% success rate. The data comes
from McCabe and Halstead measure. There is a total of 22 attributes which are used for the
prediction.
• Scope –
• Finding patterns.
• Result –
• Statistical report’s.
REFERENCES
• Trace analysis and mining for smart cities: issues, methods, and applications
By G. Pan Zhejiang Univ., Hangzhou, China G. Qi ; W. Zhang ; S. Li ; Z. Wu ; L.
T. Yang
• The home of the U.S. Government’s open data
(http://www.data.gov/dataset/college-scorecard)
• Promise Software engineering Repository
(http://promise.site.uottawa.ca/SERepository/datasets-page.html)