Using measured values at points to estimate values over a surface
A calibrated linear regression expression is created using two independent variables and a dependent variable measured at 15 points. The calibrated model is used to calculate values of the dependent variable for a network of sample points across the region. The calculated values are then interpolated using ordinary kriging, to create a continuous surface of estimated values. Finally, ESRI Spatial Analyst, Extract Values to Points, creates a point file with both measured and estimated values for error analysis. |
Using pedestrian counts and linear regression to estimate pedestrian volume on commercial streets in New York City
Six independent variables, described in the literature as possible predictors of pedestrian volume, were prepared. Of the six, only two were found to be statistically significant across the entire study area. Non-stationarity of the variables was also identified; Geographic Weighted Regression was used to limit the study area and mitigate non-stationarity. Further research requires additional count locations and additional independent variables. |
Estimating food insecurity with network analysis
In some circumstances a buffer based on Euclidean distance works well. However, social issues often require realistic measurement of distances along actual streets to support community and political dialogue. |
Working with the benefits and limitations of a harmonized world database
The Harmonized World Soil Database is downloadable as a raster at 1 Km scale and as data tables in Microsoft Access .mdb format. Four regional soils databases are harmonized in a complex process that adjusts definitions with which one may be familiar and requires careful study of the documentation. |
Identifying optimal habitat with a logistic regression model
Discrete variables (presence of humans, elk and wolves) are assigned codes, i.e. distance from humans is 1, 3 or 5, current elk territory is 1 and current wolf territory is 1. All layers are combined, via the Union overlay operation, into one layer. A new field called logit is created to hold the result of the regression equation: logit (P) = -0.847 + (1.386 * [Elk_Flag]) + (0.201 * [H_code]) + (-2.197 * [Wolf_flag]). The results are converted to probabilities that habitat is desirable. |
Using public reporting of incidents via the 311 call-in system to gauge trends
Public reports of an incident include a description of the incident and its location. Databases of public reports say at least as much about those who are willing and able to call in reports or complaints as about the phenomenon being reported. Ancillary information can round out the picture and help decision-makers and activists justify their positions and actions. |
Building a Site Suitability Model using a surface interpolated from points and binary cartographic modeling
Slope and aspect raster layers are derived from an elevation surface, and roads are buffered to a Euclidean distance of 1.5 miles. Euclidean buffering creates a raster surface of distances. The three rasters are reclassified to 0 or 1 based upon selection criteria and multiplied together. Resulting raster cells with value of 1 are potential sites. |
Identifying clusters of points and hot spots of incidence rates
Geostatistics incorporates information about the locations and relative positioning of features or their attributes into the mathematics of statistical analysis. Geostatistics responds to the fact that many phenomena are not randomly distributed and cannot be analyzed using traditional statistical methods. |
Identifying optimal habitat by combining raster & vector datasets & using boolean operations
Boolean operations can be conducted with raster layers using the raster calculator or with vector layers using geoprocessing overlay functions. Mixed data sources must be converted to one or the other for analysis. In this analysis, two datasets are converted to raster to conform with an elevation dataset. The result of analysis is a raster that is converted to vector to enable overlay operations with road layers. |