def get_results(self, path):
# all jobs should be done
res = tres.load_results(path)
mapping = {'max_steps': (-1., 'steps'), 'min_steps': (1., 'steps'),
'max_reward': (-1., 'return')}
neg, quan = mapping[self.objective]
avg, std, n_trials = tres.avg_quantity(res, quan)
avg *= neg
weights = (np.arange(len(avg)) + 1) ** 2
loss = (avg * weights).sum() / weights.sum()
print time.ctime()
print "Loss: {:.4g}".format(loss)
# use #steps/eps at the moment
return {"loss": loss,
"num_trials": n_trials[-1],
"status": hyperopt.STATUS_OK,
"std_last_mean": std[-1]}
def get_results(self, path):
# all jobs should be done
res = tres.load_results(path)
mapping = {'max_steps': (-1., 'steps'), 'min_steps': (1., 'steps'),
'max_reward': (-1., 'return')}
neg, quan = mapping[self.objective]
avg, std, n_trials = tres.avg_quantity(res, quan)
avg *= neg
weights = (np.arange(len(avg)) + 1) ** 2
loss = (avg * weights).sum() / weights.sum()
print time.ctime()
print "Loss: {:.4g}".format(loss)
# use #steps/eps at the moment
return {"loss": loss,
"num_trials": n_trials[-1],
"status": hyperopt.STATUS_OK,
"std_last_mean": std[-1]}
def f(hyperparam):
"""function to optimize by hyperopt"""
# "temporary" directory to use
full_path = os.path.join(
path, "-".join([str(v) for v in list(hyperparam.values())]))
# execute experiment
rt.run(setting,
location=full_path,
ids=list(range(1, trials_per_point + 1)),
parallelization=parallelization,
force_rerun=False,
block=True,
**hyperparam)
# all jobs should be done
res = tres.load_results(full_path)
if objective == "max_steps":
m, s, n = tres.avg_quantity(res, "steps")
val = -m
std = s[-1]
elif objective == "min_steps":
m, s, n = tres.avg_quantity(res, "steps")
val = m
std = s[-1]
elif objective == "max_reward":
m, s, n = tres.avg_quantity(res, "return")
val = -m
std = s[-1]
else:
print("unknown objective")
weights = (np.arange(len(val)) + 1)**2
loss = old_div((val * weights).sum(), weights.sum())
print(time.ctime())
print("Parameters", hyperparam)
print("Loss", loss)
# use #steps/eps at the moment
return {
"loss": loss,
"num_trials": n[-1],
"status": hyperopt.STATUS_OK,
"std_last_mean": std
}
def f(hyperparam):
"""function to optimize by hyperopt"""
# "temporary" directory to use
full_path = os.path.join(
path,
"-".join([str(v) for v in hyperparam.values()]))
# execute experiment
rt.run(setting, location=full_path, ids=range(1, trials_per_point + 1),
parallelization=parallelization, force_rerun=False, block=True, **hyperparam)
# all jobs should be done
res = tres.load_results(full_path)
if objective == "max_steps":
m, s, n = tres.avg_quantity(res, "steps")
val = -m
std = s[-1]
elif objective == "min_steps":
m, s, n = tres.avg_quantity(res, "steps")
val = m
std = s[-1]
elif objective == "max_reward":
m, s, n = tres.avg_quantity(res, "return")
val = -m
std = s[-1]
else:
print "unknown objective"
weights = (np.arange(len(val)) + 1) ** 2
loss = (val * weights).sum() / weights.sum()
print time.ctime()
print "Parameters", hyperparam
print "Loss", loss
# use #steps/eps at the moment
return {"loss": loss,
"num_trials": n[-1],
"status": hyperopt.STATUS_OK,
"std_last_mean": std}
def variable_MER_AUC(var, mp="9x9-2Path0.txt"):
#LOAD ALL RESULTS
results_agent_eps = get_all_result_paths("./TwoPathExperiments/" + var + "/" + mp + "/")
##LOAD AS VALUE -> SEED -> RESULT(dict)
tempres = dict([(x_to_y[var][int(path[-2:])], load_results(path)) for path in results_agent_eps])
##Refactor dictionary into -> SEED -> VALUE -> RESULT(dict)
seed_based_results = defaultdict(dict)
for val, total_results in tempres.items():
for seed, result in total_results.items():
seed_based_results[seed][val] = result
seed_based_AUC = defaultdict(dict)
for seed, val_res in seed_based_results.items():
for val, result in val_res.items():
seed_based_AUC[seed][val] = rt.get_AUC(result['return'])
for i, run in seed_based_AUC.items():
run_values = sorted(run.items(), key=lambda x: x[0])
keys, values = zip(*run_values)
seed_based_AUC[i] = {"Values": list(keys), "AUCS": list(values)}
return seed_based_AUC
