liut_gis_jan22_brief
TRANSCRIPT
GIS PROGRESS BRIEFTianyuan Liu
Jan 22 2016
Outline
■ Cluster Presentation (for Annotation Purpose)
■ Probability surfaces
■ Spatial weighted overlay (distance to TRO + density)
CLUSTER PRESENTATION
Aggregate Points
■ Generate polygons to enclose points that shows clustered patterns
– Tool: Cartography Tools/Generalization/Aggregate Points
– Simplify the presentation of clusters
■ Use the polygons to intersect with existing building footprints
– Potentially identify the buildings where the person spends long time
■ Caveats:
– Oversimplify the cluster
■ Cluster of 3 points or more
Aggregate Points
Distance=50m
Aggregate Points
Distance=30m
Aggregate Points
Distance=10m
Intersection with building footprintsUsing distance=10m as an example
Intersection with building footprintsFootprint data
Map Partition
■ Generate grids based on the number of features
– # of points>500
– # of points>1,000
– # of points>10,000
■ Shape size of the grids and intensity of cluster is negatively related
– Smaller grids indicates more intense cluster
■ Select the shape with smallest size and intersect with building footprint
#>500
#>10000
Building selection
#>500
Building selection
#>1000
Building selection
#>10000
PROBABILITY SURFACE
Intra-polation
■ Empirical Bayesian Kriging
– Integrate the proximal points together
– Collect the points=create z-value for calculation
– Predict the total number of points in the raster cell
■ Kernel Density (cont’d)
– Original points layer
– Kernel density + reclassify
– The raster cell need further specification
EBK
Estimating the total counts based on the
integrated points
Kernel density
SPATIAL WEIGHT OVERLAY
Spatial Weight Overlay (testing)
■ Using multiple factors to calculate the weights of each raster cell
– Euclidean distance to TROs 1(furthest) -5(nearest)
– Point density (potentially smoking events) 1(most sparse)-6(most clustered)
– Other factors
■ Caveats
– The weights need to be adjusted based on the importance of the factors
Raster cells with highest weights
TRO EXPOSURE EVALUATION
Total count of numbers within the geometric buffer
■ Using TRO buffers to intersect the original data points
– 30m buffer
– 30-50m buffer ring
– 50-100m buffer ring
■ Rank the TROs by the total number of points fall in the three buffer (ring)
■ Most exposed TROs and the distribution of activity points within the buffer
Total Count of Points Falling in the Buffers
69126
2782
0
500
1000
1500
2000
2500
3000
0_30 30_50 50_100
buffer
2466
134 91 52 41 22 20 17 13 130
500
1000
1500
2000
2500
3000
88 188 18 87 28 29 111 241 38 214
sum
Distribution of the # of points
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
88 188 18 87 28 29 111 241 38 214
% of points falling in the three buffer zones
0_30buffer 30_50buffer 50_100buffer
2% 3%
95%
%
0_30buffer 30_50buffer 50_100buffer
Most exposed TROs
TRO_ID 0_30buffer 30_50buffer 50_100buffer final name
88 7 17 2442 bull market
188 1 0 133 Walmart
18 5 55 31 Sunshine BP
87 4 3 45 Fast Food Mart
28 3 17 21 Han Dee Hugos 76
29 3 0 19 Carolina Food Mart
111 0 0 20 Academy Quick Stop
241 11 5 1 0
38 9 1 3 Stop 1 food mart
214 8 0 5 Walgreens
TRO 88 & 87