SlideShare una empresa de Scribd logo
1 de 33
Descargar para leer sin conexión
VARIABLE NEIGHBORHOOD
PREDICTION OF TEMPORAL
COLLECTIVE PROFILES
Presentation for EuroIoTA ’16
Speaker: Keun-Woo Lim
Telecom Paristech
24-11-2016
Brief Overview
 What do we do in this work?
 Analysis of temporal collective profiles (time-series)
 Use of mobile datasets (cellular, Wi-Fi)
 Real–time & Lightweight prediction (online prediction)
 What do we try to achieve?
 Better prediction accuracy
 Lower computational complexity
 Better application & use case
Contents
Contents  Introduction
 Methodology
 Prediction
 Outlier Detection
 Future Work
Introduction
Temporal collective profiles
 Representation of data that aggregate behavior of
group of individuals – over time
 Can be categorized in various ways
 “Daily Profiles”
What are collected?
 Basic telephone calls and SMS?
 However, we want to focus on more specific matters
 Specific application data
 Usage of Internet service
Why do we analyze these data?
 For “online network analysis”
 Real-time prediction of the near-future
 Recognition of sudden changes/outliers
 Timely adaptation
 Use cases
 Resource allocation
 Traffic handling
 Social behavior
Requirements
 Low computational complexity
 Lightweight prediction methods
 Accuracy
 Still have to be accurate
Dataset
 Cellular mobile dataset
 1-week data from 90 lacs in Paris
 More than 500 daily profiles
 Wi-Fi cloud dataset
 122 days (March 1st to June 30th, 2014)
 60 million URL connection logs
(Top 20 mobile applications)
Methodology
What should we do with daily profiles?
 Daily profiles can be:
 Very similar to each other (same day, location, etc.)
 Very different too (outlier, events)
 We use methods to calculate similarity
 Cluster similar profiles
 Distinguish different profiles
Previous work (Offline analysis)1
 Utilization of clustering methods (UPGMA)
 With similarity comparison techniques (DTW, quantiles)
 Not ideal in online data analysis
 Clustering may take long time (𝑂(𝑀2 𝑁3)with DTW)
1K. Lim, S. Secci, L. Tabourier, B. Tebbani, “Characterizing and predicting mobile application usage,”
https://hal.archives-ouvertes.fr/hal-01345824/document
Profile similarity
 We use two examples of similarity measures
(M values in a time-series)
 Euclidean distance (ED) = Θ(M)
 Dynamic time warping (DTW) = Θ(M2)
 For specific dataset containing N profiles,
 ED = Θ(N2M)
 DTW = Θ(N2M2)
to compare all with each other
Weighted graph representation
 Using similarity measures, we acquire a graph
structure of neighbors
 E.g., if ED is used, lower value = more similar
Filtering paths
 Filter neighbors with high distance
 Depending on the value of α, the number of neighbors
change for all profiles
Visualization of graph structure
 Example graph structure for ED – cellular dataset
Variable Neighborhood Prediction
(VNP)
Principle of VNP
 For a new day 𝑥 𝑛(𝑡), we configure
 𝑡0 = 0, 𝑡1 = 0~24, 𝑡2 = 24 (hour)
 Objective
 Observation period = 𝑥 𝑛 𝑡0, 𝑡1
 Create a temporal profile to predict 𝑥 𝑛 𝑡1, 𝑡2
 Find 𝑥 from the observation period
 The closest profile 𝑥, in 𝑥 𝑡0, 𝑡1 and 𝑥 𝑛 𝑡0, 𝑡1
Find the neighbors
 Using closest neighbor 𝑥, we find the group of
neighbors 𝑁 𝑛 to be used for prediction
 For any other profile y of the training set,
 𝑦 ∈ 𝑁𝑛 𝑖𝑖𝑓
𝑠 𝑥 𝑛 𝑡0, 𝑡1 , 𝑦 𝑡0, 𝑡1 ≤ 𝑎 ∙ 𝑠 𝑥 𝑛 𝑡0, 𝑡1 , 𝑥 𝑡0, 𝑡1
Creating the prediction profile
 Using 𝑁 𝑛, formulate the prediction
 𝑥 𝑛 𝑡 =
σ 𝑦∈𝑁 𝑛
𝑠(𝑥 𝑛,𝑦)∙𝑦(𝑡)
σ 𝑦∈𝑁 𝑛
𝑠(𝑥 𝑛,𝑦)
 Simply put, it is the weighted average over the profiles
of its neighborhood
Training Parameter 𝑎
 𝑎 can be tuned to select the optimal number of
neighbors
 Variable neighborhood search to find the 𝑎 that yields
the highest accuracy over time
 E.g. 1.0 < 𝑎 < 10.0
 Drawbacks
 Increase in complexity (recalculate for each 𝑎)
Calculating multiple 𝑡1
 For a more fine-grained prediction, multiple 𝑡1 can
be used in one day
 Repetition of the VNP (e.g. per-hour analysis)
Handling Complexity - VNP
 Computation of calculating neighborhood of target
day per 𝑎 :
 ED = Θ(NM)
 DTW = Θ(NM2)
 Depending on N, this can be large in practice
 Also, in case of multiple 𝑡1 analysis, large M can
also impact
Handling Complexity - Graph
 Can be heavy
 ED = Θ(N2M)
 DTW = Θ(N2M2)
 Luckily, graph representation is only updated once per day
 Although, needed for various M in case of multiple 𝑡1 analysis
 Also, space partitioning can be used to reduce time
 Via Kd-tree
 This can reduce complexity of ED to Θ(log(N)M)
Prediction Analysis
Prediction accuracy analysis
 Prediction through relative error, defined as
 𝜀 =
σ 𝑡=𝑡1
𝑡2 𝑥 𝑛 𝑡 − ෣𝑥 𝑛 𝑡
2
σ 𝑡=𝑡1
𝑡2 𝑥 𝑛 𝑡
2
 Comparison with closest neighbor ( 𝑎 =1), UPGMA
 𝑡1 = 12
cellular data - ED cellular data - DTW
Effect of changing 𝑡1
 Per-hour analysis
 The length of observation period may also effect the performance
of prediction
cellular data - ED cellular data - DTW
Time consumption
 The required time can be acceptable for both methods in a
per-hour analysis.
 However, need caution for DTW when many profiles are used
cellular data - ED cellular data - DTW
Distribution of α
 The distribution of optimal α is focused in range [1,2], allowing
us to easily limit the range of α
 Distribution of neighbors is heterogeneous, but most are < 20
Conclusion & Future work
Conclusion & Future work
 We have proposed a methodology for online
prediction of mobile time-series datasets
 Acceptable time for our current dataset
 Can be used for other time-series datasets in various
IoT environment
 Further studies include
 Testing in a bigger scale dataset
Any Questions?
Appendix – Wi-Fi data prediction
Wifi data - ED Wifi data - DTW

Más contenido relacionado

La actualidad más candente

K-means Clustering
K-means ClusteringK-means Clustering
K-means ClusteringAnna Fensel
 
Customer Segmentation using Clustering
Customer Segmentation using ClusteringCustomer Segmentation using Clustering
Customer Segmentation using ClusteringDessy Amirudin
 
Proximity Detection in Distributed Simulation of Wireless Mobile Systems
Proximity Detection in Distributed Simulation of Wireless Mobile SystemsProximity Detection in Distributed Simulation of Wireless Mobile Systems
Proximity Detection in Distributed Simulation of Wireless Mobile SystemsGabriele D'Angelo
 
K-Means clustring @jax
K-Means clustring @jaxK-Means clustring @jax
K-Means clustring @jaxAjay Iet
 
Grid based method & model based clustering method
Grid based method & model based clustering methodGrid based method & model based clustering method
Grid based method & model based clustering methodrajshreemuthiah
 
K means Clustering
K means ClusteringK means Clustering
K means ClusteringEdureka!
 
★Mean shift a_robust_approach_to_feature_space_analysis
★Mean shift a_robust_approach_to_feature_space_analysis★Mean shift a_robust_approach_to_feature_space_analysis
★Mean shift a_robust_approach_to_feature_space_analysisirisshicat
 
CC282 Unsupervised Learning (Clustering) Lecture 7 slides for ...
CC282 Unsupervised Learning (Clustering) Lecture 7 slides for ...CC282 Unsupervised Learning (Clustering) Lecture 7 slides for ...
CC282 Unsupervised Learning (Clustering) Lecture 7 slides for ...butest
 
Enhance The K Means Algorithm On Spatial Dataset
Enhance The K Means Algorithm On Spatial DatasetEnhance The K Means Algorithm On Spatial Dataset
Enhance The K Means Algorithm On Spatial DatasetAlaaZ
 
3.6 constraint based cluster analysis
3.6 constraint based cluster analysis3.6 constraint based cluster analysis
3.6 constraint based cluster analysisKrish_ver2
 
"k-means-clustering" presentation @ Papers We Love Bucharest
"k-means-clustering" presentation @ Papers We Love Bucharest"k-means-clustering" presentation @ Papers We Love Bucharest
"k-means-clustering" presentation @ Papers We Love BucharestAdrian Florea
 
K means and dbscan
K means and dbscanK means and dbscan
K means and dbscanYan Xu
 
Types of clustering and different types of clustering algorithms
Types of clustering and different types of clustering algorithmsTypes of clustering and different types of clustering algorithms
Types of clustering and different types of clustering algorithmsPrashanth Guntal
 
New Approach for K-mean and K-medoids Algorithm
New Approach for K-mean and K-medoids AlgorithmNew Approach for K-mean and K-medoids Algorithm
New Approach for K-mean and K-medoids AlgorithmEditor IJCATR
 
Modelling Accessibility Performance in LTE networks, An Analytics Methodology
Modelling Accessibility Performance in LTE networks, An Analytics MethodologyModelling Accessibility Performance in LTE networks, An Analytics Methodology
Modelling Accessibility Performance in LTE networks, An Analytics Methodologyalien_gmx
 
K means clustering algorithm
K means clustering algorithmK means clustering algorithm
K means clustering algorithmDarshak Mehta
 
Unsupervised Learning
Unsupervised LearningUnsupervised Learning
Unsupervised LearningAlia Hamwi
 
3.2 partitioning methods
3.2 partitioning methods3.2 partitioning methods
3.2 partitioning methodsKrish_ver2
 

La actualidad más candente (20)

K-means Clustering
K-means ClusteringK-means Clustering
K-means Clustering
 
Customer Segmentation using Clustering
Customer Segmentation using ClusteringCustomer Segmentation using Clustering
Customer Segmentation using Clustering
 
Proximity Detection in Distributed Simulation of Wireless Mobile Systems
Proximity Detection in Distributed Simulation of Wireless Mobile SystemsProximity Detection in Distributed Simulation of Wireless Mobile Systems
Proximity Detection in Distributed Simulation of Wireless Mobile Systems
 
K-Means clustring @jax
K-Means clustring @jaxK-Means clustring @jax
K-Means clustring @jax
 
Grid based method & model based clustering method
Grid based method & model based clustering methodGrid based method & model based clustering method
Grid based method & model based clustering method
 
K means Clustering
K means ClusteringK means Clustering
K means Clustering
 
★Mean shift a_robust_approach_to_feature_space_analysis
★Mean shift a_robust_approach_to_feature_space_analysis★Mean shift a_robust_approach_to_feature_space_analysis
★Mean shift a_robust_approach_to_feature_space_analysis
 
CC282 Unsupervised Learning (Clustering) Lecture 7 slides for ...
CC282 Unsupervised Learning (Clustering) Lecture 7 slides for ...CC282 Unsupervised Learning (Clustering) Lecture 7 slides for ...
CC282 Unsupervised Learning (Clustering) Lecture 7 slides for ...
 
Enhance The K Means Algorithm On Spatial Dataset
Enhance The K Means Algorithm On Spatial DatasetEnhance The K Means Algorithm On Spatial Dataset
Enhance The K Means Algorithm On Spatial Dataset
 
3.6 constraint based cluster analysis
3.6 constraint based cluster analysis3.6 constraint based cluster analysis
3.6 constraint based cluster analysis
 
"k-means-clustering" presentation @ Papers We Love Bucharest
"k-means-clustering" presentation @ Papers We Love Bucharest"k-means-clustering" presentation @ Papers We Love Bucharest
"k-means-clustering" presentation @ Papers We Love Bucharest
 
K means and dbscan
K means and dbscanK means and dbscan
K means and dbscan
 
Types of clustering and different types of clustering algorithms
Types of clustering and different types of clustering algorithmsTypes of clustering and different types of clustering algorithms
Types of clustering and different types of clustering algorithms
 
New Approach for K-mean and K-medoids Algorithm
New Approach for K-mean and K-medoids AlgorithmNew Approach for K-mean and K-medoids Algorithm
New Approach for K-mean and K-medoids Algorithm
 
Modelling Accessibility Performance in LTE networks, An Analytics Methodology
Modelling Accessibility Performance in LTE networks, An Analytics MethodologyModelling Accessibility Performance in LTE networks, An Analytics Methodology
Modelling Accessibility Performance in LTE networks, An Analytics Methodology
 
Clique
Clique Clique
Clique
 
K means clustering algorithm
K means clustering algorithmK means clustering algorithm
K means clustering algorithm
 
Unsupervised Learning
Unsupervised LearningUnsupervised Learning
Unsupervised Learning
 
3.2 partitioning methods
3.2 partitioning methods3.2 partitioning methods
3.2 partitioning methods
 
Kmeans
KmeansKmeans
Kmeans
 

Destacado

Hybrid neural networks for time series learning by Tian Guo, EPFL, Switzerland
Hybrid neural networks for time series learning by Tian Guo,  EPFL, SwitzerlandHybrid neural networks for time series learning by Tian Guo,  EPFL, Switzerland
Hybrid neural networks for time series learning by Tian Guo, EPFL, SwitzerlandEuroIoTa
 
Data analytics for monitoring IoT infrastructures by G.Madhusudan, Orange Labs
Data analytics for monitoring IoT infrastructures by G.Madhusudan, Orange LabsData analytics for monitoring IoT infrastructures by G.Madhusudan, Orange Labs
Data analytics for monitoring IoT infrastructures by G.Madhusudan, Orange LabsEuroIoTa
 
OpenML.org: Networked Science and IoT Data Streams by Jan van Rijn, Universit...
OpenML.org: Networked Science and IoT Data Streams by Jan van Rijn, Universit...OpenML.org: Networked Science and IoT Data Streams by Jan van Rijn, Universit...
OpenML.org: Networked Science and IoT Data Streams by Jan van Rijn, Universit...EuroIoTa
 
5G: a revolution or an evolution for IoT by Merouane DEBBAH, Huawei
5G: a revolution or an evolution for IoT by Merouane DEBBAH, Huawei5G: a revolution or an evolution for IoT by Merouane DEBBAH, Huawei
5G: a revolution or an evolution for IoT by Merouane DEBBAH, HuaweiEuroIoTa
 
Real-Time Big Data Stream Analytics
Real-Time Big Data Stream AnalyticsReal-Time Big Data Stream Analytics
Real-Time Big Data Stream AnalyticsAlbert Bifet
 
Grid Analytics Europe 2016: "Open for Business", April 2016
Grid Analytics Europe 2016: "Open for Business", April 2016Grid Analytics Europe 2016: "Open for Business", April 2016
Grid Analytics Europe 2016: "Open for Business", April 2016OMNETRIC
 
Big data/Data Mining/IoT/Smart City
Big data/Data Mining/IoT/Smart CityBig data/Data Mining/IoT/Smart City
Big data/Data Mining/IoT/Smart Cityrashed sharif
 
Benefiting from Big Data - A New Approach for the Telecom Industry
Benefiting from Big Data - A New Approach for the Telecom Industry  Benefiting from Big Data - A New Approach for the Telecom Industry
Benefiting from Big Data - A New Approach for the Telecom Industry Persontyle
 
Predictive Analytics in Telecommunication
Predictive Analytics in TelecommunicationPredictive Analytics in Telecommunication
Predictive Analytics in TelecommunicationRising Media Ltd.
 

Destacado (9)

Hybrid neural networks for time series learning by Tian Guo, EPFL, Switzerland
Hybrid neural networks for time series learning by Tian Guo,  EPFL, SwitzerlandHybrid neural networks for time series learning by Tian Guo,  EPFL, Switzerland
Hybrid neural networks for time series learning by Tian Guo, EPFL, Switzerland
 
Data analytics for monitoring IoT infrastructures by G.Madhusudan, Orange Labs
Data analytics for monitoring IoT infrastructures by G.Madhusudan, Orange LabsData analytics for monitoring IoT infrastructures by G.Madhusudan, Orange Labs
Data analytics for monitoring IoT infrastructures by G.Madhusudan, Orange Labs
 
OpenML.org: Networked Science and IoT Data Streams by Jan van Rijn, Universit...
OpenML.org: Networked Science and IoT Data Streams by Jan van Rijn, Universit...OpenML.org: Networked Science and IoT Data Streams by Jan van Rijn, Universit...
OpenML.org: Networked Science and IoT Data Streams by Jan van Rijn, Universit...
 
5G: a revolution or an evolution for IoT by Merouane DEBBAH, Huawei
5G: a revolution or an evolution for IoT by Merouane DEBBAH, Huawei5G: a revolution or an evolution for IoT by Merouane DEBBAH, Huawei
5G: a revolution or an evolution for IoT by Merouane DEBBAH, Huawei
 
Real-Time Big Data Stream Analytics
Real-Time Big Data Stream AnalyticsReal-Time Big Data Stream Analytics
Real-Time Big Data Stream Analytics
 
Grid Analytics Europe 2016: "Open for Business", April 2016
Grid Analytics Europe 2016: "Open for Business", April 2016Grid Analytics Europe 2016: "Open for Business", April 2016
Grid Analytics Europe 2016: "Open for Business", April 2016
 
Big data/Data Mining/IoT/Smart City
Big data/Data Mining/IoT/Smart CityBig data/Data Mining/IoT/Smart City
Big data/Data Mining/IoT/Smart City
 
Benefiting from Big Data - A New Approach for the Telecom Industry
Benefiting from Big Data - A New Approach for the Telecom Industry  Benefiting from Big Data - A New Approach for the Telecom Industry
Benefiting from Big Data - A New Approach for the Telecom Industry
 
Predictive Analytics in Telecommunication
Predictive Analytics in TelecommunicationPredictive Analytics in Telecommunication
Predictive Analytics in Telecommunication
 

Similar a Variable Neighborhood Prediction of Temporal Collective Profiles

Data Accuracy Models under Spatio - Temporal Correlation with Adaptive Strate...
Data Accuracy Models under Spatio - Temporal Correlation with Adaptive Strate...Data Accuracy Models under Spatio - Temporal Correlation with Adaptive Strate...
Data Accuracy Models under Spatio - Temporal Correlation with Adaptive Strate...IDES Editor
 
IMPROVED NEURAL NETWORK PREDICTION PERFORMANCES OF ELECTRICITY DEMAND: MODIFY...
IMPROVED NEURAL NETWORK PREDICTION PERFORMANCES OF ELECTRICITY DEMAND: MODIFY...IMPROVED NEURAL NETWORK PREDICTION PERFORMANCES OF ELECTRICITY DEMAND: MODIFY...
IMPROVED NEURAL NETWORK PREDICTION PERFORMANCES OF ELECTRICITY DEMAND: MODIFY...csandit
 
Combination of Similarity Measures for Time Series Classification using Genet...
Combination of Similarity Measures for Time Series Classification using Genet...Combination of Similarity Measures for Time Series Classification using Genet...
Combination of Similarity Measures for Time Series Classification using Genet...Deepti Dohare
 
Estimation of global solar radiation by using machine learning methods
Estimation of global solar radiation by using machine learning methodsEstimation of global solar radiation by using machine learning methods
Estimation of global solar radiation by using machine learning methodsmehmet şahin
 
DCE: A NOVEL DELAY CORRELATION MEASUREMENT FOR TOMOGRAPHY WITH PASSIVE REAL...
DCE: A NOVEL DELAY CORRELATION  MEASUREMENT FOR TOMOGRAPHY WITH PASSIVE  REAL...DCE: A NOVEL DELAY CORRELATION  MEASUREMENT FOR TOMOGRAPHY WITH PASSIVE  REAL...
DCE: A NOVEL DELAY CORRELATION MEASUREMENT FOR TOMOGRAPHY WITH PASSIVE REAL...ijdpsjournal
 
USING LEARNING AUTOMATA AND GENETIC ALGORITHMS TO IMPROVE THE QUALITY OF SERV...
USING LEARNING AUTOMATA AND GENETIC ALGORITHMS TO IMPROVE THE QUALITY OF SERV...USING LEARNING AUTOMATA AND GENETIC ALGORITHMS TO IMPROVE THE QUALITY OF SERV...
USING LEARNING AUTOMATA AND GENETIC ALGORITHMS TO IMPROVE THE QUALITY OF SERV...IJCSEA Journal
 
Data mining projects topics for java and dot net
Data mining projects topics for java and dot netData mining projects topics for java and dot net
Data mining projects topics for java and dot netredpel dot com
 
Poster_Reseau_Neurones_Journees_2013
Poster_Reseau_Neurones_Journees_2013Poster_Reseau_Neurones_Journees_2013
Poster_Reseau_Neurones_Journees_2013Pedro Lopes
 
FAST ALGORITHMS FOR UNSUPERVISED LEARNING IN LARGE DATA SETS
FAST ALGORITHMS FOR UNSUPERVISED LEARNING IN LARGE DATA SETSFAST ALGORITHMS FOR UNSUPERVISED LEARNING IN LARGE DATA SETS
FAST ALGORITHMS FOR UNSUPERVISED LEARNING IN LARGE DATA SETScsandit
 
Scheduling Using Multi Objective Genetic Algorithm
Scheduling Using Multi Objective Genetic AlgorithmScheduling Using Multi Objective Genetic Algorithm
Scheduling Using Multi Objective Genetic Algorithmiosrjce
 
LIDAR POINT CLOUD CLASSIFICATION USING EXPECTATION MAXIMIZATION ALGORITHM
LIDAR POINT CLOUD CLASSIFICATION USING EXPECTATION MAXIMIZATION ALGORITHMLIDAR POINT CLOUD CLASSIFICATION USING EXPECTATION MAXIMIZATION ALGORITHM
LIDAR POINT CLOUD CLASSIFICATION USING EXPECTATION MAXIMIZATION ALGORITHMijnlc
 
Lidar Point Cloud Classification Using Expectation Maximization Algorithm
Lidar Point Cloud Classification Using Expectation Maximization AlgorithmLidar Point Cloud Classification Using Expectation Maximization Algorithm
Lidar Point Cloud Classification Using Expectation Maximization AlgorithmAIRCC Publishing Corporation
 
IMPROVING SCHEDULING OF DATA TRANSMISSION IN TDMA SYSTEMS
IMPROVING SCHEDULING OF DATA TRANSMISSION IN TDMA SYSTEMSIMPROVING SCHEDULING OF DATA TRANSMISSION IN TDMA SYSTEMS
IMPROVING SCHEDULING OF DATA TRANSMISSION IN TDMA SYSTEMScsandit
 
IMPROVING SCHEDULING OF DATA TRANSMISSION IN TDMA SYSTEMS
IMPROVING SCHEDULING OF DATA TRANSMISSION IN TDMA SYSTEMSIMPROVING SCHEDULING OF DATA TRANSMISSION IN TDMA SYSTEMS
IMPROVING SCHEDULING OF DATA TRANSMISSION IN TDMA SYSTEMScscpconf
 
Accurate indoor positioning system based on modify nearest point technique
Accurate indoor positioning system based on modify nearest point techniqueAccurate indoor positioning system based on modify nearest point technique
Accurate indoor positioning system based on modify nearest point techniqueIJECEIAES
 
Data collection in multi application sharing wireless sensor networks
Data collection in multi application sharing wireless sensor networksData collection in multi application sharing wireless sensor networks
Data collection in multi application sharing wireless sensor networksPvrtechnologies Nellore
 
Performance of Matching Algorithmsfor Signal Approximation
Performance of Matching Algorithmsfor Signal ApproximationPerformance of Matching Algorithmsfor Signal Approximation
Performance of Matching Algorithmsfor Signal Approximationiosrjce
 

Similar a Variable Neighborhood Prediction of Temporal Collective Profiles (20)

Data Accuracy Models under Spatio - Temporal Correlation with Adaptive Strate...
Data Accuracy Models under Spatio - Temporal Correlation with Adaptive Strate...Data Accuracy Models under Spatio - Temporal Correlation with Adaptive Strate...
Data Accuracy Models under Spatio - Temporal Correlation with Adaptive Strate...
 
IMPROVED NEURAL NETWORK PREDICTION PERFORMANCES OF ELECTRICITY DEMAND: MODIFY...
IMPROVED NEURAL NETWORK PREDICTION PERFORMANCES OF ELECTRICITY DEMAND: MODIFY...IMPROVED NEURAL NETWORK PREDICTION PERFORMANCES OF ELECTRICITY DEMAND: MODIFY...
IMPROVED NEURAL NETWORK PREDICTION PERFORMANCES OF ELECTRICITY DEMAND: MODIFY...
 
Combination of Similarity Measures for Time Series Classification using Genet...
Combination of Similarity Measures for Time Series Classification using Genet...Combination of Similarity Measures for Time Series Classification using Genet...
Combination of Similarity Measures for Time Series Classification using Genet...
 
Estimation of global solar radiation by using machine learning methods
Estimation of global solar radiation by using machine learning methodsEstimation of global solar radiation by using machine learning methods
Estimation of global solar radiation by using machine learning methods
 
DCE: A NOVEL DELAY CORRELATION MEASUREMENT FOR TOMOGRAPHY WITH PASSIVE REAL...
DCE: A NOVEL DELAY CORRELATION  MEASUREMENT FOR TOMOGRAPHY WITH PASSIVE  REAL...DCE: A NOVEL DELAY CORRELATION  MEASUREMENT FOR TOMOGRAPHY WITH PASSIVE  REAL...
DCE: A NOVEL DELAY CORRELATION MEASUREMENT FOR TOMOGRAPHY WITH PASSIVE REAL...
 
USING LEARNING AUTOMATA AND GENETIC ALGORITHMS TO IMPROVE THE QUALITY OF SERV...
USING LEARNING AUTOMATA AND GENETIC ALGORITHMS TO IMPROVE THE QUALITY OF SERV...USING LEARNING AUTOMATA AND GENETIC ALGORITHMS TO IMPROVE THE QUALITY OF SERV...
USING LEARNING AUTOMATA AND GENETIC ALGORITHMS TO IMPROVE THE QUALITY OF SERV...
 
Data mining projects topics for java and dot net
Data mining projects topics for java and dot netData mining projects topics for java and dot net
Data mining projects topics for java and dot net
 
Poster_Reseau_Neurones_Journees_2013
Poster_Reseau_Neurones_Journees_2013Poster_Reseau_Neurones_Journees_2013
Poster_Reseau_Neurones_Journees_2013
 
FAST ALGORITHMS FOR UNSUPERVISED LEARNING IN LARGE DATA SETS
FAST ALGORITHMS FOR UNSUPERVISED LEARNING IN LARGE DATA SETSFAST ALGORITHMS FOR UNSUPERVISED LEARNING IN LARGE DATA SETS
FAST ALGORITHMS FOR UNSUPERVISED LEARNING IN LARGE DATA SETS
 
08039246
0803924608039246
08039246
 
M017327378
M017327378M017327378
M017327378
 
Scheduling Using Multi Objective Genetic Algorithm
Scheduling Using Multi Objective Genetic AlgorithmScheduling Using Multi Objective Genetic Algorithm
Scheduling Using Multi Objective Genetic Algorithm
 
L010628894
L010628894L010628894
L010628894
 
LIDAR POINT CLOUD CLASSIFICATION USING EXPECTATION MAXIMIZATION ALGORITHM
LIDAR POINT CLOUD CLASSIFICATION USING EXPECTATION MAXIMIZATION ALGORITHMLIDAR POINT CLOUD CLASSIFICATION USING EXPECTATION MAXIMIZATION ALGORITHM
LIDAR POINT CLOUD CLASSIFICATION USING EXPECTATION MAXIMIZATION ALGORITHM
 
Lidar Point Cloud Classification Using Expectation Maximization Algorithm
Lidar Point Cloud Classification Using Expectation Maximization AlgorithmLidar Point Cloud Classification Using Expectation Maximization Algorithm
Lidar Point Cloud Classification Using Expectation Maximization Algorithm
 
IMPROVING SCHEDULING OF DATA TRANSMISSION IN TDMA SYSTEMS
IMPROVING SCHEDULING OF DATA TRANSMISSION IN TDMA SYSTEMSIMPROVING SCHEDULING OF DATA TRANSMISSION IN TDMA SYSTEMS
IMPROVING SCHEDULING OF DATA TRANSMISSION IN TDMA SYSTEMS
 
IMPROVING SCHEDULING OF DATA TRANSMISSION IN TDMA SYSTEMS
IMPROVING SCHEDULING OF DATA TRANSMISSION IN TDMA SYSTEMSIMPROVING SCHEDULING OF DATA TRANSMISSION IN TDMA SYSTEMS
IMPROVING SCHEDULING OF DATA TRANSMISSION IN TDMA SYSTEMS
 
Accurate indoor positioning system based on modify nearest point technique
Accurate indoor positioning system based on modify nearest point techniqueAccurate indoor positioning system based on modify nearest point technique
Accurate indoor positioning system based on modify nearest point technique
 
Data collection in multi application sharing wireless sensor networks
Data collection in multi application sharing wireless sensor networksData collection in multi application sharing wireless sensor networks
Data collection in multi application sharing wireless sensor networks
 
Performance of Matching Algorithmsfor Signal Approximation
Performance of Matching Algorithmsfor Signal ApproximationPerformance of Matching Algorithmsfor Signal Approximation
Performance of Matching Algorithmsfor Signal Approximation
 

Último

Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...
Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...
Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...dajasot375
 
dokumen.tips_chapter-4-transient-heat-conduction-mehmet-kanoglu.ppt
dokumen.tips_chapter-4-transient-heat-conduction-mehmet-kanoglu.pptdokumen.tips_chapter-4-transient-heat-conduction-mehmet-kanoglu.ppt
dokumen.tips_chapter-4-transient-heat-conduction-mehmet-kanoglu.pptSonatrach
 
DBA Basics: Getting Started with Performance Tuning.pdf
DBA Basics: Getting Started with Performance Tuning.pdfDBA Basics: Getting Started with Performance Tuning.pdf
DBA Basics: Getting Started with Performance Tuning.pdfJohn Sterrett
 
RadioAdProWritingCinderellabyButleri.pdf
RadioAdProWritingCinderellabyButleri.pdfRadioAdProWritingCinderellabyButleri.pdf
RadioAdProWritingCinderellabyButleri.pdfgstagge
 
Heart Disease Classification Report: A Data Analysis Project
Heart Disease Classification Report: A Data Analysis ProjectHeart Disease Classification Report: A Data Analysis Project
Heart Disease Classification Report: A Data Analysis ProjectBoston Institute of Analytics
 
How we prevented account sharing with MFA
How we prevented account sharing with MFAHow we prevented account sharing with MFA
How we prevented account sharing with MFAAndrei Kaleshka
 
Saket, (-DELHI )+91-9654467111-(=)CHEAP Call Girls in Escorts Service Saket C...
Saket, (-DELHI )+91-9654467111-(=)CHEAP Call Girls in Escorts Service Saket C...Saket, (-DELHI )+91-9654467111-(=)CHEAP Call Girls in Escorts Service Saket C...
Saket, (-DELHI )+91-9654467111-(=)CHEAP Call Girls in Escorts Service Saket C...Sapana Sha
 
专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改
专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改
专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改yuu sss
 
INTERNSHIP ON PURBASHA COMPOSITE TEX LTD
INTERNSHIP ON PURBASHA COMPOSITE TEX LTDINTERNSHIP ON PURBASHA COMPOSITE TEX LTD
INTERNSHIP ON PURBASHA COMPOSITE TEX LTDRafezzaman
 
9654467111 Call Girls In Munirka Hotel And Home Service
9654467111 Call Girls In Munirka Hotel And Home Service9654467111 Call Girls In Munirka Hotel And Home Service
9654467111 Call Girls In Munirka Hotel And Home ServiceSapana Sha
 
Customer Service Analytics - Make Sense of All Your Data.pptx
Customer Service Analytics - Make Sense of All Your Data.pptxCustomer Service Analytics - Make Sense of All Your Data.pptx
Customer Service Analytics - Make Sense of All Your Data.pptxEmmanuel Dauda
 
Easter Eggs From Star Wars and in cars 1 and 2
Easter Eggs From Star Wars and in cars 1 and 2Easter Eggs From Star Wars and in cars 1 and 2
Easter Eggs From Star Wars and in cars 1 and 217djon017
 
Advanced Machine Learning for Business Professionals
Advanced Machine Learning for Business ProfessionalsAdvanced Machine Learning for Business Professionals
Advanced Machine Learning for Business ProfessionalsVICTOR MAESTRE RAMIREZ
 
Call Girls In Dwarka 9654467111 Escorts Service
Call Girls In Dwarka 9654467111 Escorts ServiceCall Girls In Dwarka 9654467111 Escorts Service
Call Girls In Dwarka 9654467111 Escorts ServiceSapana Sha
 
Call Us ➥97111√47426🤳Call Girls in Aerocity (Delhi NCR)
Call Us ➥97111√47426🤳Call Girls in Aerocity (Delhi NCR)Call Us ➥97111√47426🤳Call Girls in Aerocity (Delhi NCR)
Call Us ➥97111√47426🤳Call Girls in Aerocity (Delhi NCR)jennyeacort
 
办理学位证纽约大学毕业证(NYU毕业证书)原版一比一
办理学位证纽约大学毕业证(NYU毕业证书)原版一比一办理学位证纽约大学毕业证(NYU毕业证书)原版一比一
办理学位证纽约大学毕业证(NYU毕业证书)原版一比一fhwihughh
 
办理学位证中佛罗里达大学毕业证,UCF成绩单原版一比一
办理学位证中佛罗里达大学毕业证,UCF成绩单原版一比一办理学位证中佛罗里达大学毕业证,UCF成绩单原版一比一
办理学位证中佛罗里达大学毕业证,UCF成绩单原版一比一F sss
 
Predictive Analysis for Loan Default Presentation : Data Analysis Project PPT
Predictive Analysis for Loan Default  Presentation : Data Analysis Project PPTPredictive Analysis for Loan Default  Presentation : Data Analysis Project PPT
Predictive Analysis for Loan Default Presentation : Data Analysis Project PPTBoston Institute of Analytics
 

Último (20)

Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...
Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...
Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...
 
dokumen.tips_chapter-4-transient-heat-conduction-mehmet-kanoglu.ppt
dokumen.tips_chapter-4-transient-heat-conduction-mehmet-kanoglu.pptdokumen.tips_chapter-4-transient-heat-conduction-mehmet-kanoglu.ppt
dokumen.tips_chapter-4-transient-heat-conduction-mehmet-kanoglu.ppt
 
DBA Basics: Getting Started with Performance Tuning.pdf
DBA Basics: Getting Started with Performance Tuning.pdfDBA Basics: Getting Started with Performance Tuning.pdf
DBA Basics: Getting Started with Performance Tuning.pdf
 
RadioAdProWritingCinderellabyButleri.pdf
RadioAdProWritingCinderellabyButleri.pdfRadioAdProWritingCinderellabyButleri.pdf
RadioAdProWritingCinderellabyButleri.pdf
 
Heart Disease Classification Report: A Data Analysis Project
Heart Disease Classification Report: A Data Analysis ProjectHeart Disease Classification Report: A Data Analysis Project
Heart Disease Classification Report: A Data Analysis Project
 
How we prevented account sharing with MFA
How we prevented account sharing with MFAHow we prevented account sharing with MFA
How we prevented account sharing with MFA
 
Saket, (-DELHI )+91-9654467111-(=)CHEAP Call Girls in Escorts Service Saket C...
Saket, (-DELHI )+91-9654467111-(=)CHEAP Call Girls in Escorts Service Saket C...Saket, (-DELHI )+91-9654467111-(=)CHEAP Call Girls in Escorts Service Saket C...
Saket, (-DELHI )+91-9654467111-(=)CHEAP Call Girls in Escorts Service Saket C...
 
Deep Generative Learning for All - The Gen AI Hype (Spring 2024)
Deep Generative Learning for All - The Gen AI Hype (Spring 2024)Deep Generative Learning for All - The Gen AI Hype (Spring 2024)
Deep Generative Learning for All - The Gen AI Hype (Spring 2024)
 
专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改
专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改
专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改
 
E-Commerce Order PredictionShraddha Kamble.pptx
E-Commerce Order PredictionShraddha Kamble.pptxE-Commerce Order PredictionShraddha Kamble.pptx
E-Commerce Order PredictionShraddha Kamble.pptx
 
INTERNSHIP ON PURBASHA COMPOSITE TEX LTD
INTERNSHIP ON PURBASHA COMPOSITE TEX LTDINTERNSHIP ON PURBASHA COMPOSITE TEX LTD
INTERNSHIP ON PURBASHA COMPOSITE TEX LTD
 
9654467111 Call Girls In Munirka Hotel And Home Service
9654467111 Call Girls In Munirka Hotel And Home Service9654467111 Call Girls In Munirka Hotel And Home Service
9654467111 Call Girls In Munirka Hotel And Home Service
 
Customer Service Analytics - Make Sense of All Your Data.pptx
Customer Service Analytics - Make Sense of All Your Data.pptxCustomer Service Analytics - Make Sense of All Your Data.pptx
Customer Service Analytics - Make Sense of All Your Data.pptx
 
Easter Eggs From Star Wars and in cars 1 and 2
Easter Eggs From Star Wars and in cars 1 and 2Easter Eggs From Star Wars and in cars 1 and 2
Easter Eggs From Star Wars and in cars 1 and 2
 
Advanced Machine Learning for Business Professionals
Advanced Machine Learning for Business ProfessionalsAdvanced Machine Learning for Business Professionals
Advanced Machine Learning for Business Professionals
 
Call Girls In Dwarka 9654467111 Escorts Service
Call Girls In Dwarka 9654467111 Escorts ServiceCall Girls In Dwarka 9654467111 Escorts Service
Call Girls In Dwarka 9654467111 Escorts Service
 
Call Us ➥97111√47426🤳Call Girls in Aerocity (Delhi NCR)
Call Us ➥97111√47426🤳Call Girls in Aerocity (Delhi NCR)Call Us ➥97111√47426🤳Call Girls in Aerocity (Delhi NCR)
Call Us ➥97111√47426🤳Call Girls in Aerocity (Delhi NCR)
 
办理学位证纽约大学毕业证(NYU毕业证书)原版一比一
办理学位证纽约大学毕业证(NYU毕业证书)原版一比一办理学位证纽约大学毕业证(NYU毕业证书)原版一比一
办理学位证纽约大学毕业证(NYU毕业证书)原版一比一
 
办理学位证中佛罗里达大学毕业证,UCF成绩单原版一比一
办理学位证中佛罗里达大学毕业证,UCF成绩单原版一比一办理学位证中佛罗里达大学毕业证,UCF成绩单原版一比一
办理学位证中佛罗里达大学毕业证,UCF成绩单原版一比一
 
Predictive Analysis for Loan Default Presentation : Data Analysis Project PPT
Predictive Analysis for Loan Default  Presentation : Data Analysis Project PPTPredictive Analysis for Loan Default  Presentation : Data Analysis Project PPT
Predictive Analysis for Loan Default Presentation : Data Analysis Project PPT
 

Variable Neighborhood Prediction of Temporal Collective Profiles

  • 1. VARIABLE NEIGHBORHOOD PREDICTION OF TEMPORAL COLLECTIVE PROFILES Presentation for EuroIoTA ’16 Speaker: Keun-Woo Lim Telecom Paristech 24-11-2016
  • 2. Brief Overview  What do we do in this work?  Analysis of temporal collective profiles (time-series)  Use of mobile datasets (cellular, Wi-Fi)  Real–time & Lightweight prediction (online prediction)  What do we try to achieve?  Better prediction accuracy  Lower computational complexity  Better application & use case
  • 3. Contents Contents  Introduction  Methodology  Prediction  Outlier Detection  Future Work
  • 5. Temporal collective profiles  Representation of data that aggregate behavior of group of individuals – over time  Can be categorized in various ways  “Daily Profiles”
  • 6. What are collected?  Basic telephone calls and SMS?  However, we want to focus on more specific matters  Specific application data  Usage of Internet service
  • 7. Why do we analyze these data?  For “online network analysis”  Real-time prediction of the near-future  Recognition of sudden changes/outliers  Timely adaptation  Use cases  Resource allocation  Traffic handling  Social behavior
  • 8. Requirements  Low computational complexity  Lightweight prediction methods  Accuracy  Still have to be accurate
  • 9. Dataset  Cellular mobile dataset  1-week data from 90 lacs in Paris  More than 500 daily profiles  Wi-Fi cloud dataset  122 days (March 1st to June 30th, 2014)  60 million URL connection logs (Top 20 mobile applications)
  • 11. What should we do with daily profiles?  Daily profiles can be:  Very similar to each other (same day, location, etc.)  Very different too (outlier, events)  We use methods to calculate similarity  Cluster similar profiles  Distinguish different profiles
  • 12. Previous work (Offline analysis)1  Utilization of clustering methods (UPGMA)  With similarity comparison techniques (DTW, quantiles)  Not ideal in online data analysis  Clustering may take long time (𝑂(𝑀2 𝑁3)with DTW) 1K. Lim, S. Secci, L. Tabourier, B. Tebbani, “Characterizing and predicting mobile application usage,” https://hal.archives-ouvertes.fr/hal-01345824/document
  • 13. Profile similarity  We use two examples of similarity measures (M values in a time-series)  Euclidean distance (ED) = Θ(M)  Dynamic time warping (DTW) = Θ(M2)  For specific dataset containing N profiles,  ED = Θ(N2M)  DTW = Θ(N2M2) to compare all with each other
  • 14. Weighted graph representation  Using similarity measures, we acquire a graph structure of neighbors  E.g., if ED is used, lower value = more similar
  • 15. Filtering paths  Filter neighbors with high distance  Depending on the value of α, the number of neighbors change for all profiles
  • 16. Visualization of graph structure  Example graph structure for ED – cellular dataset
  • 18. Principle of VNP  For a new day 𝑥 𝑛(𝑡), we configure  𝑡0 = 0, 𝑡1 = 0~24, 𝑡2 = 24 (hour)  Objective  Observation period = 𝑥 𝑛 𝑡0, 𝑡1  Create a temporal profile to predict 𝑥 𝑛 𝑡1, 𝑡2  Find 𝑥 from the observation period  The closest profile 𝑥, in 𝑥 𝑡0, 𝑡1 and 𝑥 𝑛 𝑡0, 𝑡1
  • 19. Find the neighbors  Using closest neighbor 𝑥, we find the group of neighbors 𝑁 𝑛 to be used for prediction  For any other profile y of the training set,  𝑦 ∈ 𝑁𝑛 𝑖𝑖𝑓 𝑠 𝑥 𝑛 𝑡0, 𝑡1 , 𝑦 𝑡0, 𝑡1 ≤ 𝑎 ∙ 𝑠 𝑥 𝑛 𝑡0, 𝑡1 , 𝑥 𝑡0, 𝑡1
  • 20. Creating the prediction profile  Using 𝑁 𝑛, formulate the prediction  𝑥 𝑛 𝑡 = σ 𝑦∈𝑁 𝑛 𝑠(𝑥 𝑛,𝑦)∙𝑦(𝑡) σ 𝑦∈𝑁 𝑛 𝑠(𝑥 𝑛,𝑦)  Simply put, it is the weighted average over the profiles of its neighborhood
  • 21. Training Parameter 𝑎  𝑎 can be tuned to select the optimal number of neighbors  Variable neighborhood search to find the 𝑎 that yields the highest accuracy over time  E.g. 1.0 < 𝑎 < 10.0  Drawbacks  Increase in complexity (recalculate for each 𝑎)
  • 22. Calculating multiple 𝑡1  For a more fine-grained prediction, multiple 𝑡1 can be used in one day  Repetition of the VNP (e.g. per-hour analysis)
  • 23. Handling Complexity - VNP  Computation of calculating neighborhood of target day per 𝑎 :  ED = Θ(NM)  DTW = Θ(NM2)  Depending on N, this can be large in practice  Also, in case of multiple 𝑡1 analysis, large M can also impact
  • 24. Handling Complexity - Graph  Can be heavy  ED = Θ(N2M)  DTW = Θ(N2M2)  Luckily, graph representation is only updated once per day  Although, needed for various M in case of multiple 𝑡1 analysis  Also, space partitioning can be used to reduce time  Via Kd-tree  This can reduce complexity of ED to Θ(log(N)M)
  • 26. Prediction accuracy analysis  Prediction through relative error, defined as  𝜀 = σ 𝑡=𝑡1 𝑡2 𝑥 𝑛 𝑡 − ෣𝑥 𝑛 𝑡 2 σ 𝑡=𝑡1 𝑡2 𝑥 𝑛 𝑡 2  Comparison with closest neighbor ( 𝑎 =1), UPGMA  𝑡1 = 12 cellular data - ED cellular data - DTW
  • 27. Effect of changing 𝑡1  Per-hour analysis  The length of observation period may also effect the performance of prediction cellular data - ED cellular data - DTW
  • 28. Time consumption  The required time can be acceptable for both methods in a per-hour analysis.  However, need caution for DTW when many profiles are used cellular data - ED cellular data - DTW
  • 29. Distribution of α  The distribution of optimal α is focused in range [1,2], allowing us to easily limit the range of α  Distribution of neighbors is heterogeneous, but most are < 20
  • 31. Conclusion & Future work  We have proposed a methodology for online prediction of mobile time-series datasets  Acceptable time for our current dataset  Can be used for other time-series datasets in various IoT environment  Further studies include  Testing in a bigger scale dataset
  • 33. Appendix – Wi-Fi data prediction Wifi data - ED Wifi data - DTW