def test_AreaAnalyst(self): aa = AreaAnalyst(self.r1, self.r1) raster = aa.getChangeMap() band = raster.getBand(1) assert_array_equal(band, self.r1r1) # Masked raster aa = AreaAnalyst(self.r2, self.r2) raster = aa.getChangeMap() band = raster.getBand(1) assert_array_equal(band, self.r2r2)
def test_AreaAnalyst(self): aa = AreaAnalyst(self.r1, self.r1) raster = aa.getChangeMap() band = raster.getBand(1) assert_array_equal(band, self.r1r1) # Masked raster aa = AreaAnalyst(self.r2, self.r2) raster = aa.getChangeMap() band = raster.getBand(1) assert_array_equal(band, self.r2r2)
def sim(self): """ Make 1 iteracion of simulation. """ # TODO: eleminate AreaAnalyst.getChangeMap() from the process transition = self.crosstable.getCrosstable() prediction = self.getPrediction() state = self.getState() new_state = state.getBand(1).copy() # New states (the result of simulation) will be stored there. analyst = AreaAnalyst(state, prediction) classes = analyst.classes changes = analyst.getChangeMap().getBand(1) # Make transition between classes according to # number of moved pixel in crosstable self.rangeChanged.emit(self.tr("Simulation process %p%"), len(classes) ** 2 - len(classes)) for initClass in classes: for finalClass in classes: if initClass == finalClass: continue # TODO: Calculate number of pixels to be moved via TransitoionMatrix and state raster n = transition.getTransition( initClass, finalClass ) # Number of pixels to be moved (constant count now). # Find n appropriate places for transition initClass -> finalClass class_code = analyst.encode(initClass, finalClass) places = changes == class_code # Array of places where transitions initClass -> finalClass are occured placesCount = np.sum(places) if placesCount < n: self.logMessage.emit( self.tr("There are more transitions in the transition matrix, then the model have found") ) n = placesCount confidence = self.getConfidence().getBand(1) confidence = ( confidence * places ) # The higher is number in cell, the higer is probability of transition in the cell indices = [] for i in range(n): index = np.unravel_index( confidence.argmax(), confidence.shape ) # Select the cell with biggest probability indices.append(index) confidence[index] = -1 # Mark the cell to prevent second selection # Now "indices" contains indices of the appropriate places, # make transition initClass -> finalClass for index in indices: new_state[index] = finalClass self.updateProgress.emit() result = Raster() result.create([new_state], state.getGeodata()) self.state = result self.updatePrediction(result) self.processFinished.emit()
def main(initRaster, finalRaster, factors): print 'Start Reading Init Data...', clock() initRaster = Raster(initRaster) finalRaster = Raster(finalRaster) factors = [Raster(rasterName) for rasterName in factors] print 'Finish Reading Init Data', clock(), '\n' print "Start Making CrossTable...", clock() crosstab = CrossTableManager(initRaster, finalRaster) #print crosstab.getTransitionStat() print "Finish Making CrossTable", clock(), '\n' # Create and Train Analyst print 'Start creating AreaAnalyst...', clock() analyst = AreaAnalyst(initRaster, finalRaster) print 'Finish creating AreaAnalyst ...', clock(), '\n' print 'Start Making Change Map...', clock() analyst = AreaAnalyst(initRaster, finalRaster) changeMap = analyst.getChangeMap() print 'Finish Making Change Map', clock(), '\n' #~ # Create and Train ANN Model model = MlpManager(ns=1) model.createMlp(initRaster, factors, changeMap, [10]) print 'Start Setting MLP Trainig Data...', clock() model.setTrainingData(initRaster, factors, changeMap, mode='Stratified', samples=1000) print 'Finish Setting Trainig Data', clock(), '\n' print 'Start MLP Training...', clock() model.train(20, valPercent=20) print 'Finish Trainig', clock(), '\n' # print 'Start ANN Prediction...', clock() # predict = model.getPrediction(initRaster, factors, calcTransitions=True) # confidence = model.getConfidence() # potentials = model.getTransitionPotentials() #~ # Create and Train LR Model #~ model = LR(ns=0) #~ print 'Start Setting LR Trainig Data...', clock() #~ model.setState(initRaster) #~ model.setFactors(factors) #~ model.setOutput(changeMap) #~ model.setMode('Stratified') #~ model.setSamples(100) #~ model.setTrainingData() #~ print 'Finish Setting Trainig Data', clock(), '\n' #~ print 'Start LR Training...', clock() #~ model.train() #~ print 'Finish Trainig', clock(), '\n' #~ #~ print 'Start LR Prediction...', clock() #~ predict = model.getPrediction(initRaster, factors, calcTransitions=True) #~ print 'Finish LR Prediction...', clock(), '\n' # Create and Train WoE Model # print 'Start creating AreaAnalyst...', clock() # analyst = AreaAnalyst(initRaster, finalRaster) # print 'Finish creating AreaAnalyst ...', clock(), '\n' # print 'Start creating WoE model...', clock() # bins = {0: [[1000, 3000]], 1: [[200, 500, 1500]]} # model = WoeManager(factors, analyst, bins= bins) # model.train() # print 'Finish creating WoE model...', clock(), '\n' #~ # Create and Train MCE Model #~ print 'Start creating MCE model...', clock() #~ matrix = [ #~ [1, 6], #~ [1.0/6, 1] #~ ] #~ model = MCE(factors, matrix, 2, 3, analyst) #~ print 'Finish creating MCE model...', clock(), '\n' # predict = model.getPrediction(initRaster, factors, calcTransitions=True) # confidence = model.getConfidence() # potentials = model.getTransitionPotentials() # filename = 'predict.tif' # confname = 'confidence.tif' # trans_prefix='trans_' # try: # predict.save(filename) # confidence.save(confname) # if potentials != None: # for k,v in potentials.iteritems(): # map = v.save(trans_prefix+str(k) + '.tif') # finally: # os.remove(filename) # #pass # print 'Finish Saving...', clock(), '\n' # simulation print 'Start Simulation...', clock() simulator = Simulator(initRaster, factors, model, crosstab) # Make 1 cycle of simulation: simulator.setIterationCount(1) simulator.simN() monteCarloSim = simulator.getState() # Result of MonteCarlo simulation errors = simulator.errorMap(finalRaster) # Risk class validation riskFunct = simulator.getConfidence() # Risk function try: monteCarloSim.save('simulation_result.tiff') errors.save('risk_validation.tiff') riskFunct.save('risk_func.tiff') finally: pass # os.remove('simulation_result.tiff') # os.remove('risk_validation.tiff') # os.remove('risk_func.tiff') print 'Finish Simulation', clock(), '\n' print 'Done', clock()
def main(initRaster, finalRaster, factors): print 'Start Reading Init Data...', clock() initRaster = Raster(initRaster) finalRaster = Raster(finalRaster) factors = [Raster(rasterName) for rasterName in factors] print 'Finish Reading Init Data', clock(), '\n' print "Start Making CrossTable...", clock() crosstab = CrossTableManager(initRaster, finalRaster) #print crosstab.getTransitionStat() print "Finish Making CrossTable", clock(), '\n' # Create and Train Analyst print 'Start creating AreaAnalyst...', clock() analyst = AreaAnalyst(initRaster, finalRaster) print 'Finish creating AreaAnalyst ...', clock(), '\n' print 'Start Making Change Map...', clock() analyst = AreaAnalyst(initRaster,finalRaster) changeMap = analyst.getChangeMap() print 'Finish Making Change Map', clock(), '\n' #~ # Create and Train ANN Model model = MlpManager(ns=1) model.createMlp(initRaster, factors, changeMap, [10]) print 'Start Setting MLP Trainig Data...', clock() model.setTrainingData(initRaster, factors, changeMap, mode='Stratified', samples=1000) print 'Finish Setting Trainig Data', clock(), '\n' print 'Start MLP Training...', clock() model.train(20, valPercent=20) print 'Finish Trainig', clock(), '\n' # print 'Start ANN Prediction...', clock() # predict = model.getPrediction(initRaster, factors, calcTransitions=True) # confidence = model.getConfidence() # potentials = model.getTransitionPotentials() #~ # Create and Train LR Model #~ model = LR(ns=0) #~ print 'Start Setting LR Trainig Data...', clock() #~ model.setState(initRaster) #~ model.setFactors(factors) #~ model.setOutput(changeMap) #~ model.setMode('Stratified') #~ model.setSamples(100) #~ model.setTrainingData() #~ print 'Finish Setting Trainig Data', clock(), '\n' #~ print 'Start LR Training...', clock() #~ model.train() #~ print 'Finish Trainig', clock(), '\n' #~ #~ print 'Start LR Prediction...', clock() #~ predict = model.getPrediction(initRaster, factors, calcTransitions=True) #~ print 'Finish LR Prediction...', clock(), '\n' # Create and Train WoE Model # print 'Start creating AreaAnalyst...', clock() # analyst = AreaAnalyst(initRaster, finalRaster) # print 'Finish creating AreaAnalyst ...', clock(), '\n' # print 'Start creating WoE model...', clock() # bins = {0: [[1000, 3000]], 1: [[200, 500, 1500]]} # model = WoeManager(factors, analyst, bins= bins) # model.train() # print 'Finish creating WoE model...', clock(), '\n' #~ # Create and Train MCE Model #~ print 'Start creating MCE model...', clock() #~ matrix = [ #~ [1, 6], #~ [1.0/6, 1] #~ ] #~ model = MCE(factors, matrix, 2, 3, analyst) #~ print 'Finish creating MCE model...', clock(), '\n' # predict = model.getPrediction(initRaster, factors, calcTransitions=True) # confidence = model.getConfidence() # potentials = model.getTransitionPotentials() # filename = 'predict.tif' # confname = 'confidence.tif' # trans_prefix='trans_' # try: # predict.save(filename) # confidence.save(confname) # if potentials != None: # for k,v in potentials.iteritems(): # map = v.save(trans_prefix+str(k) + '.tif') # finally: # os.remove(filename) # #pass # print 'Finish Saving...', clock(), '\n' # simulation print 'Start Simulation...', clock() simulator = Simulator(initRaster, factors, model, crosstab) # Make 1 cycle of simulation: simulator.setIterationCount(1) simulator.simN() monteCarloSim = simulator.getState() # Result of MonteCarlo simulation errors = simulator.errorMap(finalRaster) # Risk class validation riskFunct = simulator.getConfidence() # Risk function try: monteCarloSim.save('simulation_result.tiff') errors.save('risk_validation.tiff') riskFunct.save('risk_func.tiff') finally: pass # os.remove('simulation_result.tiff') # os.remove('risk_validation.tiff') # os.remove('risk_func.tiff') print 'Finish Simulation', clock(), '\n' print 'Done', clock()