def nearlyGreedyStrat(counts, locPrior, likelihoodsObj, dag, skipProbs):
dlocs = [i for i in xrange(len(counts)) if counts[i][1]]
(currEntropy, entropyResults,
findProbs) = entropiesFast(counts, locPrior, likelihoodsObj, dag)
(next, nextE) = findMin(entropyResults)
if len(dlocs):
# if there is a detection, calculate the expected entropy after making another observation
# there and then making a 'greedy' observation.
dloc = dlocs[-1]
(t, d) = counts[dloc]
dcounts = copy.copy(counts)
tcounts = copy.copy(counts)
dcounts[dloc] = (t, d + 1)
tcounts[dloc] = (t + 1, d)
(currEntropyD, entropyResultsD,
findProbsD) = entropiesFast(dcounts, locPrior, likelihoodsObj, dag)
(currEntropyT, entropyResultsT,
findProbsT) = entropiesFast(tcounts, locPrior, likelihoodsObj, dag)
(nextD, nextED) = findMin(entropyResultsD)
(nextT, nextET) = findMin(entropyResultsT)
expectedEntropy = findProbs[dloc] * nextED + (1 -
findProbs[dloc]) * nextET
# print "c %1.2f n %1.02f c-n %1.04f c-e %1.04f fp %1.02f nf %1.02f nt %1.02f" %(currEntropy,nextE,currEntropy-nextE,currEntropy-expectedEntropy,findProbs[dloc],nextED,nextET)
if (currEntropy - nextE) < (currEntropy - expectedEntropy) / 2.0:
return dloc
else:
return next
else:
return next
def nearlyGreedyStrat(counts,locPrior,likelihoodsObj,dag,skipProbs):
dlocs=[i for i in xrange(len(counts)) if counts[i][1]]
(currEntropy,entropyResults,findProbs)=entropiesFast(counts,locPrior,likelihoodsObj,dag)
(next,nextE)=findMin(entropyResults)
if len(dlocs):
# if there is a detection, calculate the expected entropy after making another observation
# there and then making a 'greedy' observation.
dloc=dlocs[-1]
(t,d)=counts[dloc]
dcounts=copy.copy(counts)
tcounts=copy.copy(counts)
dcounts[dloc]=(t,d+1)
tcounts[dloc]=(t+1,d)
(currEntropyD,entropyResultsD,findProbsD)=entropiesFast(dcounts,locPrior,likelihoodsObj,dag)
(currEntropyT,entropyResultsT,findProbsT)=entropiesFast(tcounts,locPrior,likelihoodsObj,dag)
(nextD,nextED)=findMin(entropyResultsD)
(nextT,nextET)=findMin(entropyResultsT)
expectedEntropy=findProbs[dloc]*nextED+(1-findProbs[dloc])*nextET
# print "c %1.2f n %1.02f c-n %1.04f c-e %1.04f fp %1.02f nf %1.02f nt %1.02f" %(currEntropy,nextE,currEntropy-nextE,currEntropy-expectedEntropy,findProbs[dloc],nextED,nextET)
if (currEntropy-nextE)<(currEntropy-expectedEntropy)/2.0:
return dloc
else:
return next
else:
return next
def greedyStrat(counts,locPrior,likelihoodsObj,dag,skipProbs):
(currEntropy,entropyResults,findProbs)=entropiesFast(counts,locPrior,likelihoodsObj,dag)
# test where expected entropy is smallest
expectedGain = [(currEntropy-entropyResults[i])*(numberType.one-skipProbs[i]) for
i in xrange(len(entropyResults))]
(next,nextp)=findMax(expectedGain)
return next
def greedyStrat(counts, locPrior, likelihoodsObj, dag, skipProbs):
(currEntropy, entropyResults,
findProbs) = entropiesFast(counts, locPrior, likelihoodsObj, dag)
# test where expected entropy is smallest
expectedGain = [
(currEntropy - entropyResults[i]) * (numberType.one - skipProbs[i])
for i in xrange(len(entropyResults))
]
(next, nextp) = findMax(expectedGain)
return next
import testCases
random.seed(1)
mult=100
out=file("data/N_T.singleRate.csv","w")
for n in range(50,1000,50):
d=testCases.testDag(n,False)
for i in range(mult):
(counts,locPrior)=testCases.randomEntropyData(i+1,n,d,False,True,10)
start=time.clock()
junk=evidence.entropiesFast(counts,locPrior,likelihoods.singleRateCalc,d)
end=time.clock()
out.write("%d,%f\n" % (n,(end-start)))
print n
out.close()
#
# You should have received a copy of the GNU General Public License
# along with BBChop. If not, see <http://www.gnu.org/licenses/>.
import evidence
import time
import random
import likelihoods
import testCases
random.seed(1)
mult = 100
out = file("data/N_T.singleRate.csv", "w")
for n in range(50, 1000, 50):
d = testCases.testDag(n, False)
for i in range(mult):
(counts,
locPrior) = testCases.randomEntropyData(i + 1, n, d, False, True, 10)
start = time.clock()
junk = evidence.entropiesFast(counts, locPrior,
likelihoods.singleRateCalc, d)
end = time.clock()
out.write("%d,%f\n" % (n, (end - start)))
print n
out.close()
