Geovisualization and Geostatistics: A Concept for the Numerical and Visual Analysis of Geographic Mass Data Julia Gonschorek, Lucia Tyrallová, - University of Potsdam
Download as PPS, PDF1 like1,832 views
The document discusses a concept for analyzing and visualizing geographic mass data through geovisualization and geostatistics. It presents methods for extracting spatiotemporal information from large emergency services databases in Cologne, Germany, including line plots, box plots, heatmaps, and kernel density estimation maps. The analysis identified temporal and location-based clusters in emergency cases. Future work will focus on using these exploratory techniques and geovisualization methods to better understand emergency response patterns and support prevention and planning efforts.
![ICCSA Brazil 2012
Magisterarbeit 6/12
6/??
2. Spatio-Temporal Analysis for Civil Security (2)
Box-and-Whisker-Plots to show differences in varaiances:
R-Code (without months “January” and “February”):
emergency <- read.csv(“c:tempintern07.csv”,header=T, sep=“;”)
chisq.test(emergency[1:7,4:13])
Pearson's Chi-squared test
data: emergency[1:7, 4:13]
X-squared = 409.496, df = 54, p-value < 2.2e-16
qchisq(0.95,54)
72.15322
The non-parametric χ²- Test for testing dependency validates the observation.
The correlation between “Day of the Week” and “Months” is highly significant.
@gonschorek ∙ university of potsdam](https://image.slidesharecdn.com/juliaiccsa-120713092102-phpapp02/85/Geovisualization-and-Geostatistics-A-Concept-for-the-Numerical-and-Visual-Analysis-of-Geographic-Mass-Data-Julia-Gonschorek-Lucia-Tyrallova-University-of-Potsdam-6-320.jpg)
Geovisualization and Geostatistics: A Concept for the Numerical and Visual Analysis of Geographic Mass Data Julia Gonschorek, Lucia Tyrallová, - University of Potsdam
- 1. Magisterarbeit 1/?? 12 International Conference on Computational Science and Its Applications th (ICCSA2012 in Salvador da Bahia/Brazil) Session GeoAnMod-3 on Monday June 18, 2012 Geovisualization and Geostatistics: A Concept for the Numerical and Visual Analysis of Geographic Mass Data Julia Gonschorek | Geoinformation Research Group | University of Potsdam Co-Author: Lucia Tyrallová | Geoinformation Research Group | University of Potsdam
- 2. ICCSA Brazil 2012 Magisterarbeit 2/12 2/?? Outline 1. Preface and Motivation 2. Spatio-Temporal Analysis for Civil Security 3. Summary and Future Plans @gonschorek ∙ university of potsdam
- 3. ICCSA Brazil 2012 Magisterarbeit 3/12 3/?? 1. Preface and Motivation (1) Increasing availability of (mass-) data and need for specific information complex computational analysis tools and techniques Highly dimensional data needs to be analysed rapidly to discover relationships, clusters and trends Scientific visualization offers a wide range of methods and techniques to efficiently analyze and visualize spatial and temporal data and information @gonschorek ∙ university of potsdam
- 4. ICCSA Brazil 2012 Magisterarbeit 4/12 4/?? 1. Preface and Motivation (2) @gonschorek ∙ university of potsdam
- 5. ICCSA Brazil 2012 Magisterarbeit 5/12 5/?? 2. Spatio-Temporal Analysis for Civil Security (1) Simple lineplot to visualize the temporal distribution of internistic emergencies in the City of Cologne (total number of 26,475 in 2007): @gonschorek ∙ university of potsdam
- 6. ICCSA Brazil 2012 Magisterarbeit 6/12 6/?? 2. Spatio-Temporal Analysis for Civil Security (2) Box-and-Whisker-Plots to show differences in varaiances: R-Code (without months “January” and “February”): emergency <- read.csv(“c:tempintern07.csv”,header=T, sep=“;”) chisq.test(emergency[1:7,4:13]) Pearson's Chi-squared test data: emergency[1:7, 4:13] X-squared = 409.496, df = 54, p-value < 2.2e-16 qchisq(0.95,54) 72.15322 The non-parametric χ²- Test for testing dependency validates the observation. The correlation between “Day of the Week” and “Months” is highly significant. @gonschorek ∙ university of potsdam
- 7. ICCSA Brazil 2012 Magisterarbeit 7/12 7/?? 2. Spatio-Temporal Analysis for Civil Security (3) Heatmap to detect temporal clusters: R-Code: install.packages(“gplots”) install.packages (“RColorBrewer”) library(gplots) library(RColorBrewer) x <- read.csv(“c:temp intern07.csv”, header=T, sep=“;”, row.names=1) matrix=data.matrix(x) heatmap.2(matrix, Rowv=T, Colv=T, dendrogram=c(“none”), distfun=dist, hclustfun=hclust, key=T, keysize=1, trace=“none”, density.info=c(“none”), margins=c(10,10), col=brewer.pal(10,”PiYG”)) @gonschorek ∙ university of potsdam
- 8. ICCSA Brazil 2012 Magisterarbeit 8/12 8/?? 2. Spatio-Temporal Analysis for Civil Security (4) Map of 17,000 surgery emergencies in the City of Cologne during July, 2007 and June 2008 with kernel density estimation using Epanechnikov kernel: Hotspots of surgery emergencies… on Friday on Saturday on Sunday @gonschorek ∙ university of potsdam
- 9. ICCSA Brazil 2012 Magisterarbeit 9/12 9/?? 2. Spatio-Temporal Analysis for Civil Security (5) (e) (d) Surgical Emergencies > total: 300 (c) > daytime: 6.00 – 7.00 a.m. > year: 2009 > district: Altstadt Sued (b) (a) (a) Data Source: all incoming emergency calls (b) First-order circle: inhomogeneous parts for cluster or administrative information (urban districts) (c) Second-order circle: homogeneous parts for temporal information: year (d) Third-order circle: inhomogeneous parts for temporal information: month, daytime, … @gonschorek ∙ university of potsdam (e) Fourth-order circle: inhomogeneous parts; Type of emergency case
- 10. ICCSA Brazil 2012 Magisterarbeit 10/12 10/?? 3. Summary and Future Plans (1) Methods can be used to efficiently extract spatio-temporal information from large databases It is shown how specific emergency services cluster in space and time Nearly the whole city area of Cologne was an emergency scene during the analysed perid. Especially the city centre, leisure facilities and nursing homes were emergency hot spots Combination of different explorative techniques with those from geovisualisation can check the (long-term) experience of the firefighters on different spatial scales and precision @gonschorek ∙ university of potsdam
- 11. ICCSA Brazil 2012 Magisterarbeit 11/12 11/?? 3. Summary and Future Plans (2) Methods and results are important for future explorative analyses and geovisual analytics Information and a deep understanding of specific distributions and patterns as well as (ir-) regularities of intensity are absolutely necessary for prevention measurements to be well-directed and needs based. Time-series-analysis and prognoses are suitable for the operational, strategic and tactical planning. @gonschorek ∙ university of potsdam
- 12. Magisterarbeit 12/?? Thank you for your attention! Julia Gonschorek | julia.gonschorek@uni-potsdam.de Department of Geography | University of Potsdam http://www.geographie.uni-potsdam.de/geoinformatik