nbDetectedDecision = []
exploredDecisions = {}
exploredSeqDecisions = {}
exploredLocations = {}
sizeCorrectSeqDecisions = []
sizeFailureSeqDecisions = []
for execution in data['executions']:
if 'decisions' not in execution:
continue
if execution['result']['success']:
countPassedLaps += 1
sID = ""
for decision in execution['decisions']:
lID = locationID(decision['location'])
if lID not in exploredLocations:
exploredLocations[lID] = 0
exploredLocations[lID] += 1
dID = decisionID(decision)
sID += dID
if dID not in exploredDecisions:
exploredDecisions[dID] = 0
exploredDecisions[dID] += 1
if sID not in exploredSeqDecisions:
exploredSeqDecisions[sID] = 0
if execution['result']['success']:
nbPassedSeqLaps += 1
exploredSeqDecisions[sID] += 1
def plot(bugName, bug):
output = """ \\begin{tikzpicture}
\\begin{axis}[
enlargelimits=false,
xmin=1,xmax=200,
ymin=0,ymax=100,
ytick={0,10,20,30,40,50,60,70,80,90,100},
ymajorgrids=true,
grid style=dashed,
smooth,
axis lines=left,
legend pos=outer north east,
ylabel={\\% Correct Laps},
xlabel={Nb Laps Before Max Explored Decision},
]
"""
#ylabel={Nb Laps Before Stabilization},
epsilons = sorted(bug)
for epsilon in epsilons:
if "%.1f" % float(epsilon) == "0.0":
continue
coordinates = ""
seeds = sorted(bug[epsilon])
for seed in seeds:
executionsPath = join(path, bugName, str(bug[epsilon][seed]) + ".json")
with open(executionsPath) as data_file:
data = json.load(data_file)
index = 0
countPassedLaps = 0
nbBeforeStabilization = -1
exploredDecisions = {}
exploredSeqDecisions = {}
exploredLocations = {}
learns = []
sizeCorrectSeqDecisions = []
sizeFailureSeqDecisions = []
maxExplredDecision = 0
nbLapsBeforeMaxDecision = 0
for execution in data['executions']:
if 'decisions' not in execution:
continue
if execution['result']['success']:
countPassedLaps += 1
sID = ""
for decision in execution['decisions']:
lID = locationID(decision['location'])
if lID not in exploredLocations:
exploredLocations[lID] = 0
exploredLocations[lID] += 1
dID = decisionID(decision)
sID += dID
if dID not in exploredDecisions:
exploredDecisions[dID] = 0
exploredDecisions[dID] += 1
# learning curve
learn = countPassedLaps*100/(index + 1)
learns += [learn]
if len(learns) >= 10 and nbBeforeStabilization == -1 and learn != 0:
minLearn = 100
maxLearn = 0
for learn in learns[index - 9::]:
if learn > maxLearn:
maxLearn = learn
if learn < minLearn:
minLearn = learn
diffLearn = maxLearn - minLearn
if diffLearn <= 4:
nbBeforeStabilization = index - 9
if len(exploredDecisions) > maxExplredDecision:
maxExplredDecision = len(exploredDecisions)
nbLapsBeforeMaxDecision = index + 1
index += 1
#print seed, epsilon, nbLapsBeforeMaxDecision, nbBeforeStabilization
coordinates += "(%d,%d) " % (nbLapsBeforeMaxDecision, countPassedLaps*100/len(data['executions']))
output += " \\addplot+[only marks] coordinates {%s};" % coordinates
output +=" \\addlegendentry{\\exploitationCoefficient %.1f}\n" % (1 - float(epsilon))
output += """ \end{axis}
\end{tikzpicture}"""
return output