def learn(self): #print self.cs, self.ts learning = learn.learnCurve(self.cs, self.ts) #print learning if not math.isnan(learning['r']): #if it's too huge, just set it to 1000 if learning['r'] > 5000.0: self.er = 5000.0 else: if learning['r'] <= 0: self.er = 5000.0 else: self.er = learning['r'] self.epv = learning['pv'] self.erv = learning['rv'] else: self.er = 5000.0 self.epv = 1.0 self.erv = 1.0 if not math.isnan(learning['p']): if learning['p'] < 0: self.ep = 0.0 else: self.ep = learning['p'] else: self.ep = 0.0
total = 0 maximum = 0 count = 0 worker = Worker(str(uuid.uuid1()), 0, p, r, 1, 1) worker.addNoise(0, 0.1) for i in range(0, runs): ts = [] cs = [] for j in range(0, horizon): task = tasks[j] answer = worker.doTask(task, outcomes) if answer == task: count += 1 ts.append(j + 1) cs.append(count) learning = learn.learnCurve(cs, ts) err = learning['e'] #print err #print learning['r'] if math.isnan(err): continue if err < threshold: total += j + 1 tally.append(j+1) if j + 1 > maximum: maximum = j + 1 break ys1.append(float(total) / float(runs)) ys2.append(maximum) xs = np.arange(r_start, r_end + 1, 1)
def getCurves(worker, tasks, runs, prob):
#the following two are always fixed
cqs = [] #cumulative quality
qs = [] #quality
aqs = [] #average quality
ecqs = [] #estimated by linear regression
eqs = [] #estimated by linear regression
fqs = []
###for learning
cs = []
ts = []
errs = []
count = 0
for i in range(0, len(tasks)):
task = tasks[i]
answer = worker.doTask(task)
if answer == getAnswer(task, prob):
count += 1
cqs.append(worker.getCumulativeQuality(i + 1))
qs.append(worker.getQuality())
aqs.append(float(count) / float(i + 1))
#learn
ts.append(i + 1)
cs.append(count)
learning = learn.learnCurve(cs, ts)
fake = Worker(str(uuid.uuid1()), i+1, learning['p'], learning['r'], 1, 1)
ecqs.append(fake.getCumulativeQuality(i+1))
eqs.append(fake.getQuality())
errs.append(learning['e'])
if i == 0:
fqs.append(float(count) / float(i+1))
else:
fqs.append(float(i+1)*aqs[i] - float(i)*aqs[i-1])
for j in range(0, runs - 1):
worker.reset()
count = 0
cs = []
ts = []
for i in range(0, len(tasks)):
task = tasks[i]
answer = worker.doTask(task)
if answer == getAnswer(task, prob):
count += 1
aqs[i] += float(count) / float(i + 1)
ts.append(i + 1)
cs.append(count)
learning = learn.learnCurve(cs, ts)
fake = Worker(str(uuid.uuid1()), i+1, learning['p'], learning['r'], 1, 1)
ecqs[i] += fake.getCumulativeQuality(i+1)
eqs[i] += fake.getQuality()
errs[i] += learning['e']
if i == 0:
fqs[i] += (float(count) / float(i+1))
else:
fqs[i] += (float(i+1)*aqs[i] - float(i)*aqs[i-1])
for i in range(0, len(tasks)):
aqs[i] = aqs[i] / float(runs)
ecqs[i] = ecqs[i] / float(runs)
eqs[i] = eqs[i] / float(runs)
errs[i] = errs[i] / float(runs)
fqs[i] = fqs[i] / float(runs)
result = {'cqs': cqs, 'qs': qs, 'aqs': aqs, 'ecqs': ecqs, 'eqs': eqs}
return result