pysal.model.mgwr.gwr.GWRResults

class pysal.model.mgwr.gwr.GWRResults(model, params, predy, S, CCT, w=None)[source]

Basic class including common properties for all GWR regression models

Parameters:
model : GWR object

pointer to GWR object with estimation parameters

params : array

n*k, estimated coefficients

predy : array

n*1, predicted y values

S : array

n*n, hat matrix

CCT : array

n*k, scaled variance-covariance matrix

w : array

n*1, final weight used for iteratively re-weighted least sqaures; default is None

Attributes:
model : GWR Object

points to GWR object for which parameters have been estimated

params : array

n*k, parameter estimates

predy : array

n*1, predicted value of y

y : array

n*1, dependent variable

X : array

n*k, independent variable, including constant

family : family object

underlying probability model; provides distribution-specific calculations

n : integer

number of observations

k : integer

number of independent variables

df_model : integer

model degrees of freedom

df_resid : integer

residual degrees of freedom

offset : array

n*1, the offset variable at the ith location. For Poisson model this term is often the size of the population at risk or the expected size of the outcome in spatial epidemiology; Default is None where Ni becomes 1.0 for all locations

scale : float

sigma squared used for subsequent computations

w : array

n*1, final weights from iteratively re-weighted least sqaures routine

resid_response : array

n*1, residuals of the repsonse

resid_ss : scalar

residual sum of sqaures

W : array

n*n; spatial weights for each observation from each calibration point

S : array

n*n, hat matrix

CCT : array

n*k, scaled variance-covariance matrix

ENP : scalar

effective number of parameters

tr_S : float

trace of S (hat) matrix

tr_STS : float

trace of STS matrix

y_bar : array

weighted mean of y

TSS : array

geographically weighted total sum of squares

RSS : array

geographically weighted residual sum of squares

R2 : float

R-squared for the entire model (1- RSS/TSS)

aic : float

Akaike information criterion

aicc : float

corrected Akaike information criterion to account to account for model complexity (smaller bandwidths)

bic : float

Bayesian information criterio

localR2 : array

local R square

sigma2 : float

residual variance

std_res : array

standardized residuals

bse : array

standard errors of Betas

influ : array

Influence: leading diagonal of S Matrix

CooksD : array

n*1, Cook’s D

tvalues : array

Return the t-statistic for a given parameter estimate.

adj_alpha : array

Corrected alpha (critical) values to account for multiple testing during hypothesis testing.

deviance : array

n*1, local model deviance for each calibration point

resid_deviance : array

n*1, local sum of residual deviance for each calibration point

llf : scalar

log-likelihood of the full model; see pysal.contrib.glm.family for damily-sepcific log-likelihoods

pDev : float

Local percentage of deviance accounted for.

mu : array

n*, flat one dimensional array of predicted mean response value from estimator

fit_params : dict

parameters passed into fit method to define estimation routine

predictions : array

p*1, predicted values generated by calling the GWR predict method to predict dependent variable at unsampled points ()

Methods

ENP() effective number of parameters
RSS() geographically weighted residual sum of squares
TSS() geographically weighted total sum of squares
adj_alpha() Corrected alpha (critical) values to account for multiple testing during hypothesis testing.
bse() standard errors of Betas
conf_int() Returns the confidence interval of the fitted parameters.
cooksD() Influence: leading diagonal of S Matrix
cov_params(cov[, exog_scale]) Returns scaled covariance parameters
critical_tval([alpha]) Utility function to derive the critial t-value based on given alpha that are needed for hypothesis testing
filter_tvals([critical_t, alpha]) Utility function to set tvalues with an absolute value smaller than the absolute value of the alpha (critical) value to 0.
influ() Influence: leading diagonal of S Matrix
localR2() local R square
local_collinearity() Computes several indicators of multicollinearity within a geographically weighted design matrix, including:
pDev() Local percentage of deviance accounted for.
sigma2() residual variance
spatial_variability(selector[, n_iters, seed]) Method to compute a Monte Carlo test of spatial variability for each estimated coefficient surface.
std_res() standardized residuals
summary() Print out GWR summary
tr_S() trace of S (hat) matrix
tr_STS() trace of STS matrix
tvalues() Return the t-statistic for a given parameter estimate.
use_t() bool(x) -> bool
y_bar() weighted mean of y
D2  
R2  
adj_D2  
adj_pseudoR2  
aic  
aicc  
bic  
deviance  
df_model  
df_resid  
initialize  
llf  
llnull  
normalized_cov_params  
null  
null_deviance  
pearson_chi2  
predictions  
pseudoR2  
pvalues  
resid_anscombe  
resid_deviance  
resid_pearson  
resid_response  
resid_ss  
resid_working  
scale  
__init__(model, params, predy, S, CCT, w=None)[source]

Initialize self. See help(type(self)) for accurate signature.

Methods

D2()
ENP() effective number of parameters
R2()
RSS() geographically weighted residual sum of squares
TSS() geographically weighted total sum of squares
__init__(model, params, predy, S, CCT[, w]) Initialize self.
adj_D2()
adj_alpha() Corrected alpha (critical) values to account for multiple testing during hypothesis testing.
adj_pseudoR2()
aic()
aicc()
bic()
bse() standard errors of Betas
conf_int() Returns the confidence interval of the fitted parameters.
cooksD() Influence: leading diagonal of S Matrix
cov_params(cov[, exog_scale]) Returns scaled covariance parameters
critical_tval([alpha]) Utility function to derive the critial t-value based on given alpha that are needed for hypothesis testing
deviance()
df_model()
df_resid()
filter_tvals([critical_t, alpha]) Utility function to set tvalues with an absolute value smaller than the absolute value of the alpha (critical) value to 0.
influ() Influence: leading diagonal of S Matrix
initialize(model, params, **kwd)
llf()
llnull()
localR2() local R square
local_collinearity() Computes several indicators of multicollinearity within a geographically weighted design matrix, including:
normalized_cov_params()
null()
null_deviance()
pDev() Local percentage of deviance accounted for.
pearson_chi2()
predictions()
pseudoR2()
pvalues()
resid_anscombe()
resid_deviance()
resid_pearson()
resid_response()
resid_ss()
resid_working()
scale([scale])
sigma2() residual variance
spatial_variability(selector[, n_iters, seed]) Method to compute a Monte Carlo test of spatial variability for each estimated coefficient surface.
std_res() standardized residuals
summary() Print out GWR summary
tr_S() trace of S (hat) matrix
tr_STS() trace of STS matrix
tvalues() Return the t-statistic for a given parameter estimate.
use_t() bool(x) -> bool
y_bar() weighted mean of y