Crime geography: journey to crime, circle hypothesis, hit score and risk terrain
Source:R/CrimeGeography.R
JtcDecay.RdJtcDecay: CrimeStat journey-to-crime decay functions (Levine 2013,
equations 13.14-13.20): linear \(\max(0, A + Bd)\), negative
exponential \(Ae^{-Bd}\), normal \(A\phi((d-\bar d)/S)/S\), lognormal
\(A/(d^2 S\sqrt{2\pi})\exp(-(\ln d^2 - \bar d)^2/(2S^2))\) and the
truncated negative exponential (linear to the cutoff, exponential beyond),
with CrimeStat's Baltimore County defaults. JtcSurface: the summed
likelihood over grid points (Euclidean or Manhattan distance).
JtcCalibrate: grouped-data calibration of all five functions
(equations 13.22-13.44) with residual sums of squares.
CircleHypothesis: Canter and Larkin's marauder test.
SearchCost: the hit percentage of a profile. RiskLayers:
proximity or density layers of features around cell centres.
RiskTerrain: Poisson IRLS regression of cell counts on risk layers,
relative risk values \(e^{b_k}\) and the composite relative risk
(Caplan, Kennedy and Miller 2011). Identical to the Python arm
morie.fn.crimegeo.
Usage
JtcDecay(d, fn = "negative_exponential", params = NULL)
JtcSurface(
incidents,
grid,
fn = "negative_exponential",
params = NULL,
metric = "euclidean"
)
JtcCalibrate(distances, breaks)
CircleHypothesis(crimes, home = NULL)
SearchCost(scores, home_index)
RiskLayers(cells, features, radius, kind = "proximity")
RiskTerrain(counts, layers, names = NULL, tol = 1e-10, maxit = 50L)Arguments
- d
Distances.
- fn
Decay function name.
- params
Named list overriding the default parameters.
- incidents, crimes
Two-column incident coordinates.
- grid
Two-column grid coordinates.
- metric
"euclidean"or"manhattan".- distances
Known offender journey distances.
- breaks
Distance bin breaks.
- home
Offender residence (optional).
- scores
Profile scores of the search cells.
- home_index
Index (1-based) of the home cell.
- cells
Two-column cell centres.
- features
Two-column feature locations.
- radius
Search radius.
- kind
"proximity"or"density".- counts
Crime counts per cell.
- layers
List (or matrix columns) of risk layers.
- names
Layer names.
- tol
Relative deviance tolerance.
- maxit
Maximum IRLS iterations.
References
Levine, N. (2013). Journey-to-crime estimation. Chapter 13 in CrimeStat IV: A Spatial Statistics Program for the Analysis of Crime Incident Locations, version 4.0. National Institute of Justice (NCJ 242973).
Canter, D. and Larkin, P. (1993). The environmental range of serial rapists. Journal of Environmental Psychology 13, 63-69.
Rossmo, D. K. (2000). Geographic Profiling. CRC Press.
Caplan, J. M., Kennedy, L. W. and Miller, J. (2011). Risk terrain modeling: brokering criminological theory and GIS methods for crime forecasting. Justice Quarterly 28, 360-381.