def oneModelRunWithOutput(self, allocationDictionary):
import costcov
import data
from copy import deepcopy as dcp
costCov = costcov.Costcov()
spreadsheetData = data.readSpreadsheet(self.dataSpreadsheetName, self.helper.keyList)
model, derived, params = self.helper.setupModelConstantsParameters(spreadsheetData)
costCoverageInfo = self.getCostCoverageInfo()
# run the model
modelList = []
timestepsPre = 12
for t in range(timestepsPre):
model.moveOneTimeStep()
modelThisTimeStep = dcp(model)
modelList.append(modelThisTimeStep)
# update coverages after 1 year
targetPopSize = getTargetPopSizeFromModelInstance(self.dataSpreadsheetName, self.helper.keyList, model)
newCoverages = {}
for i in range(0, len(spreadsheetData.interventionList)):
intervention = spreadsheetData.interventionList[i]
newCoverages[intervention] = costCov.function(allocationDictionary[intervention], costCoverageInfo[intervention], targetPopSize[intervention]) / targetPopSize[intervention]
model.updateCoverages(newCoverages)
for t in range(self.numModelSteps - timestepsPre):
model.moveOneTimeStep()
modelThisTimeStep = dcp(model)
modelList.append(modelThisTimeStep)
return modelList
def performCascadeOptimisation(self, optimise, MCSampleSize, filename,
cascadeValues):
import data
spreadsheetData = data.readSpreadsheet(self.dataSpreadsheetName,
self.helper.keyList)
costCoverageInfo = self.getCostCoverageInfo()
initialTargetPopSize = self.getInitialTargetPopSize()
initialAllocation = getTotalInitialAllocation(spreadsheetData,
costCoverageInfo,
initialTargetPopSize)
currentTotalBudget = sum(initialAllocation)
xmin = [0.] * len(initialAllocation)
for cascade in cascadeValues:
totalBudget = currentTotalBudget * cascade
args = {
'totalBudget': totalBudget,
'costCoverageInfo': costCoverageInfo,
'optimise': optimise,
'numModelSteps': self.numModelSteps,
'dataSpreadsheetName': self.dataSpreadsheetName,
'data': spreadsheetData
}
self.runOnce(MCSampleSize, xmin, args,
spreadsheetData.interventionList, totalBudget,
filename + str(cascade) + '.pkl')
def setUpDataModelConstantsParameters():
helperTests = helper.Helper()
testData = data.readSpreadsheet(
'input_spreadsheets/InputForCode_tests.xlsx', helperTests.keyList)
testModel, testDerived, testParams = helperTests.setupModelConstantsParameters(
testData)
return testData, testModel, testDerived, testParams, helperTests.keyList
def generateBOCVectors(self, filenameStem, regionNameList, cascadeValues,
outcome):
import pickle
import data
spreadsheetData = data.readSpreadsheet(self.dataSpreadsheetName,
self.helper.keyList)
costCoverageInfo = self.getCostCoverageInfo()
targetPopSize = self.getInitialTargetPopSize()
initialAllocation = getTotalInitialAllocation(spreadsheetData,
costCoverageInfo,
targetPopSize)
currentTotalBudget = sum(initialAllocation)
spendingVector = []
outcomeVector = []
for cascade in cascadeValues:
spendingVector.append(cascade * currentTotalBudget)
filename = filenameStem + '_cascade_' + str(outcome) + '_' + str(
cascade) + '.pkl'
infile = open(filename, 'rb')
thisAllocation = pickle.load(infile)
infile.close()
modelOutput = self.oneModelRunWithOutput(thisAllocation)
if outcome == 'deaths':
outcomeVector.append(modelOutput[self.numModelSteps -
1].getTotalCumulativeDeaths())
if outcome == 'stunting':
outcomeVector.append(
modelOutput[self.numModelSteps -
1].getCumulativeAgingOutStunted())
return spendingVector, outcomeVector
def oneModelRunWithOutput(self, allocationDictionary):
import costcov
import data
from copy import deepcopy as dcp
costCov = costcov.Costcov()
spreadsheetData = data.readSpreadsheet(self.dataSpreadsheetName,
self.helper.keyList)
model, derived, params = self.helper.setupModelConstantsParameters(
spreadsheetData)
costCoverageInfo = self.getCostCoverageInfo()
# run the model
modelList = []
timestepsPre = 12
for t in range(timestepsPre):
model.moveOneTimeStep()
modelThisTimeStep = dcp(model)
modelList.append(modelThisTimeStep)
# update coverages after 1 year
targetPopSize = getTargetPopSizeFromModelInstance(
self.dataSpreadsheetName, self.helper.keyList, model)
newCoverages = {}
for i in range(0, len(spreadsheetData.interventionList)):
intervention = spreadsheetData.interventionList[i]
newCoverages[intervention] = costCov.function(
allocationDictionary[intervention],
costCoverageInfo[intervention],
targetPopSize[intervention]) / targetPopSize[intervention]
model.updateCoverages(newCoverages)
for t in range(self.numModelSteps - timestepsPre):
model.moveOneTimeStep()
modelThisTimeStep = dcp(model)
modelList.append(modelThisTimeStep)
return modelList
def performSingleOptimisationForGivenTotalBudget(self, optimise, MCSampleSize, filename, totalBudget):
import data
spreadsheetData = data.readSpreadsheet(self.dataSpreadsheetName, self.helper.keyList)
costCoverageInfo = self.getCostCoverageInfo()
xmin = [0.] * len(spreadsheetData.interventionList)
args = {'totalBudget':totalBudget, 'costCoverageInfo':costCoverageInfo, 'optimise':optimise, 'numModelSteps':self.numModelSteps, 'dataSpreadsheetName':self.dataSpreadsheetName, 'data':spreadsheetData}
self.runOnce(MCSampleSize, xmin, args, spreadsheetData.interventionList, totalBudget, filename+'.pkl')
def performSingleOptimisation(self, optimise, MCSampleSize, filename):
import data
spreadsheetData = data.readSpreadsheet(self.dataSpreadsheetName, self.helper.keyList)
costCoverageInfo = self.getCostCoverageInfo()
initialTargetPopSize = self.getInitialTargetPopSize()
initialAllocation = getTotalInitialAllocation(spreadsheetData, costCoverageInfo, initialTargetPopSize)
totalBudget = sum(initialAllocation)
xmin = [0.] * len(initialAllocation)
args = {'totalBudget':totalBudget, 'costCoverageInfo':costCoverageInfo, 'optimise':optimise, 'numModelSteps':self.numModelSteps, 'dataSpreadsheetName':self.dataSpreadsheetName, 'data':spreadsheetData}
self.runOnce(MCSampleSize, xmin, args, spreadsheetData.interventionList, totalBudget, filename+'.pkl')
def getCostCoverageInfo(self):
import data
from copy import deepcopy as dcp
spreadsheetData = data.readSpreadsheet(self.dataSpreadsheetName, self.helper.keyList)
costCoverageInfo = {}
for intervention in spreadsheetData.interventionList:
costCoverageInfo[intervention] = {}
costCoverageInfo[intervention]['unitcost'] = dcp(spreadsheetData.costSaturation[intervention]["unit cost"])
costCoverageInfo[intervention]['saturation'] = dcp(spreadsheetData.costSaturation[intervention]["saturation coverage"])
return costCoverageInfo
def getInitialAllocationDictionary(self):
import data
spreadsheetData = data.readSpreadsheet(self.dataSpreadsheetName, self.helper.keyList)
costCoverageInfo = self.getCostCoverageInfo()
targetPopSize = self.getInitialTargetPopSize()
initialAllocation = getTotalInitialAllocation(spreadsheetData, costCoverageInfo, targetPopSize)
initialAllocationDictionary = {}
for i in range(0, len(spreadsheetData.interventionList)):
intervention = spreadsheetData.interventionList[i]
initialAllocationDictionary[intervention] = initialAllocation[i]
return initialAllocationDictionary
def performParallelCascadeOptimisation(self, optimise, MCSampleSize, filename, cascadeValues):
import data
from joblib import Parallel, delayed
spreadsheetData = data.readSpreadsheet(self.dataSpreadsheetName, self.helper.keyList)
costCoverageInfo = self.getCostCoverageInfo()
initialTargetPopSize = self.getInitialTargetPopSize()
initialAllocation = getTotalInitialAllocation(spreadsheetData, costCoverageInfo, initialTargetPopSize)
currentTotalBudget = sum(initialAllocation)
xmin = [0.] * len(initialAllocation)
# use one core per cascade value
nCores = len(cascadeValues)
Parallel(n_jobs=nCores)(delayed(self.cascadeFunc)(cascadeValue, currentTotalBudget, costCoverageInfo, optimise, MCSampleSize, xmin, filename) for cascadeValue in cascadeValues)
def getTargetPopSizeFromModelInstance(dataSpreadsheetName, keyList, model):
import data
spreadsheetData = data.readSpreadsheet(dataSpreadsheetName, keyList)
numAgeGroups = len(keyList['ages'])
targetPopSize = {}
for intervention in spreadsheetData.interventionList:
targetPopSize[intervention] = 0.
for iAge in range(numAgeGroups):
ageName = keyList['ages'][iAge]
targetPopSize[intervention] += spreadsheetData.targetPopulation[intervention][ageName] * model.listOfAgeCompartments[iAge].getTotalPopulation()
targetPopSize[intervention] += spreadsheetData.targetPopulation[intervention]['pregnant women'] * model.pregnantWomen.populationSize
return targetPopSize
def getCostCoverageInfo(self):
import data
from copy import deepcopy as dcp
spreadsheetData = data.readSpreadsheet(self.dataSpreadsheetName,
self.helper.keyList)
costCoverageInfo = {}
for intervention in spreadsheetData.interventionList:
costCoverageInfo[intervention] = {}
costCoverageInfo[intervention]['unitcost'] = dcp(
spreadsheetData.costSaturation[intervention]["unit cost"])
costCoverageInfo[intervention]['saturation'] = dcp(
spreadsheetData.costSaturation[intervention]
["saturation coverage"])
return costCoverageInfo
def getInitialAllocationDictionary(self):
import data
spreadsheetData = data.readSpreadsheet(self.dataSpreadsheetName,
self.helper.keyList)
costCoverageInfo = self.getCostCoverageInfo()
targetPopSize = self.getInitialTargetPopSize()
initialAllocation = getTotalInitialAllocation(spreadsheetData,
costCoverageInfo,
targetPopSize)
initialAllocationDictionary = {}
for i in range(0, len(spreadsheetData.interventionList)):
intervention = spreadsheetData.interventionList[i]
initialAllocationDictionary[intervention] = initialAllocation[i]
return initialAllocationDictionary
def getInitialTargetPopSize(self):
import data
spreadsheetData = data.readSpreadsheet(self.dataSpreadsheetName, self.helper.keyList)
mothers = self.helper.makePregnantWomen(spreadsheetData)
numAgeGroups = len(self.helper.keyList['ages'])
agePopSizes = self.helper.makeAgePopSizes(spreadsheetData)
targetPopSize = {}
for intervention in spreadsheetData.interventionList:
targetPopSize[intervention] = 0.
for iAge in range(numAgeGroups):
ageName = self.helper.keyList['ages'][iAge]
targetPopSize[intervention] += spreadsheetData.targetPopulation[intervention][ageName] * agePopSizes[iAge]
targetPopSize[intervention] += spreadsheetData.targetPopulation[intervention]['pregnant women'] * mothers.populationSize
return targetPopSize
def getTotalNationalBudget(self):
import optimisation
import data
regionalBudgets = []
for region in range(0, self.numRegions):
thisSpreadsheet = self.regionSpreadsheetList[region]
thisOptimisation = optimisation.Optimisation(thisSpreadsheet, self.numModelSteps)
spreadsheetData = data.readSpreadsheet(thisSpreadsheet, thisOptimisation.helper.keyList)
costCoverageInfo = thisOptimisation.getCostCoverageInfo()
initialTargetPopSize = thisOptimisation.getInitialTargetPopSize()
initialAllocation = getTotalInitialAllocation(spreadsheetData, costCoverageInfo, initialTargetPopSize)
regionTotalBudget = sum(initialAllocation)
regionalBudgets.append(regionTotalBudget)
nationalTotalBudget = sum(regionalBudgets)
return nationalTotalBudget
def getTargetPopSizeFromModelInstance(dataSpreadsheetName, keyList, model):
import data
spreadsheetData = data.readSpreadsheet(dataSpreadsheetName, keyList)
numAgeGroups = len(keyList['ages'])
targetPopSize = {}
for intervention in spreadsheetData.interventionList:
targetPopSize[intervention] = 0.
for iAge in range(numAgeGroups):
ageName = keyList['ages'][iAge]
targetPopSize[intervention] += spreadsheetData.targetPopulation[
intervention][ageName] * model.listOfAgeCompartments[
iAge].getTotalPopulation()
targetPopSize[
intervention] += spreadsheetData.targetPopulation[intervention][
'pregnant women'] * model.pregnantWomen.populationSize
return targetPopSize
def getTotalNationalBudget(self):
import optimisation
import data
regionalBudgets = []
for region in range(0, self.numRegions):
thisSpreadsheet = self.regionSpreadsheetList[region]
thisOptimisation = optimisation.Optimisation(
thisSpreadsheet, self.numModelSteps)
spreadsheetData = data.readSpreadsheet(
thisSpreadsheet, thisOptimisation.helper.keyList)
costCoverageInfo = thisOptimisation.getCostCoverageInfo()
initialTargetPopSize = thisOptimisation.getInitialTargetPopSize()
initialAllocation = getTotalInitialAllocation(
spreadsheetData, costCoverageInfo, initialTargetPopSize)
regionTotalBudget = sum(initialAllocation)
regionalBudgets.append(regionTotalBudget)
nationalTotalBudget = sum(regionalBudgets)
return nationalTotalBudget
def performParallelCascadeOptimisation(self, optimise, MCSampleSize,
filename, cascadeValues):
import data
from joblib import Parallel, delayed
spreadsheetData = data.readSpreadsheet(self.dataSpreadsheetName,
self.helper.keyList)
costCoverageInfo = self.getCostCoverageInfo()
initialTargetPopSize = self.getInitialTargetPopSize()
initialAllocation = getTotalInitialAllocation(spreadsheetData,
costCoverageInfo,
initialTargetPopSize)
currentTotalBudget = sum(initialAllocation)
xmin = [0.] * len(initialAllocation)
# use one core per cascade value
nCores = len(cascadeValues)
Parallel(n_jobs=nCores)(delayed(self.cascadeFunc)(
cascadeValue, currentTotalBudget, costCoverageInfo, optimise,
MCSampleSize, xmin, filename) for cascadeValue in cascadeValues)
def getInitialTargetPopSize(self):
import data
spreadsheetData = data.readSpreadsheet(self.dataSpreadsheetName,
self.helper.keyList)
mothers = self.helper.makePregnantWomen(spreadsheetData)
numAgeGroups = len(self.helper.keyList['ages'])
agePopSizes = self.helper.makeAgePopSizes(spreadsheetData)
targetPopSize = {}
for intervention in spreadsheetData.interventionList:
targetPopSize[intervention] = 0.
for iAge in range(numAgeGroups):
ageName = self.helper.keyList['ages'][iAge]
targetPopSize[
intervention] += spreadsheetData.targetPopulation[
intervention][ageName] * agePopSizes[iAge]
targetPopSize[intervention] += spreadsheetData.targetPopulation[
intervention]['pregnant women'] * mothers.populationSize
return targetPopSize
def performSingleOptimisationForGivenTotalBudget(self, optimise,
MCSampleSize, filename,
totalBudget):
import data
spreadsheetData = data.readSpreadsheet(self.dataSpreadsheetName,
self.helper.keyList)
costCoverageInfo = self.getCostCoverageInfo()
xmin = [0.] * len(spreadsheetData.interventionList)
args = {
'totalBudget': totalBudget,
'costCoverageInfo': costCoverageInfo,
'optimise': optimise,
'numModelSteps': self.numModelSteps,
'dataSpreadsheetName': self.dataSpreadsheetName,
'data': spreadsheetData
}
self.runOnce(MCSampleSize, xmin, args,
spreadsheetData.interventionList, totalBudget,
filename + '.pkl')
def generateBOCVectors(self, filenameStem, regionNameList, cascadeValues, outcome):
import pickle
import data
spreadsheetData = data.readSpreadsheet(self.dataSpreadsheetName, self.helper.keyList)
costCoverageInfo = self.getCostCoverageInfo()
targetPopSize = self.getInitialTargetPopSize()
initialAllocation = getTotalInitialAllocation(spreadsheetData, costCoverageInfo, targetPopSize)
currentTotalBudget = sum(initialAllocation)
spendingVector = []
outcomeVector = []
for cascade in cascadeValues:
spendingVector.append(cascade * currentTotalBudget)
filename = filenameStem + '_cascade_' + str(outcome) + '_' + str(cascade)+'.pkl'
infile = open(filename, 'rb')
thisAllocation = pickle.load(infile)
infile.close()
modelOutput = self.oneModelRunWithOutput(thisAllocation)
if outcome == 'deaths':
outcomeVector.append(modelOutput[self.numModelSteps-1].getTotalCumulativeDeaths())
if outcome == 'stunting':
outcomeVector.append(modelOutput[self.numModelSteps-1].getCumulativeAgingOutStunted())
return spendingVector, outcomeVector
import data
import helper
import output
import costcov
helper = helper.Helper()
costCov = costcov.Costcov()
country = 'Bangladesh'
startYear = 2016
version = '1604'
dataFilename = '../input_spreadsheets/%s/InputForCode_%s.xlsx' % (country,
country)
#dataFilename = '../input_spreadsheets/%s/Input_%s_%i_%s.xlsx'%(country, country, startYear, version)
inputData = data.readSpreadsheet(dataFilename, helper.keyList)
numAgeGroups = len(helper.keyList['ages'])
numsteps = 180
oldCoverages = {}
costCovParams = {}
print "Baseline coverage of: "
for intervention in inputData.interventionList:
print "%s = %g" % (intervention, inputData.coverage[intervention])
oldCoverages[intervention] = inputData.coverage[intervention]
# calculate coverage (%)
costCovParams[intervention] = {}
costCovParams[intervention]['unitcost'] = dcp(
inputData.costSaturation[intervention]["unit cost"])
costCovParams[intervention]['saturation'] = dcp(
def setUpDataModelConstantsParameters():
helperTests = helper.Helper()
testData = data.readSpreadsheet('input_spreadsheets/InputForCode_tests.xlsx', helperTests.keyList)
testModel, testDerived, testParams = helperTests.setupModelConstantsParameters(testData)
return testData, testModel, testDerived, testParams, helperTests.keyList
import os, sys moduleDir = os.path.join(os.path.dirname(__file__), "..") sys.path.append(moduleDir) import data import helper import output helper = helper.Helper() country = "Bangladesh" startYear = 2000 dataFilename = "../input_spreadsheets/%s/validation/Input_%s_%i_1604.xlsx" % (country, country, startYear) inputData = data.readSpreadsheet(dataFilename, helper.keyList) numAgeGroups = len(helper.keyList["ages"]) year1 = 3871841 year2 = 3731124 year3 = 3645815 year4 = 3567786 year5 = 3491964 agePopSizes = [year1 / 12.0, year1 * 5.0 / 12.0, year1 * 6.0 / 12.0, year2, year3 + year4 + year5] numsteps = 192 timespan = helper.keyList["timestep"] * float(numsteps) # endYear = startYear + int(timespan) numYears = int(timespan) yearList = list(range(startYear, startYear + numYears)) # initialise
