D_in, D_out = expert_data["X"].shape[-1], expert_data["y"].shape[-1]
env = gym.make(args["env"])
model_fn = getattr(models, args["model_fn"])
model_dir = args["model_dir"]
model = model_fn(D_in, D_out, model_dir=model_dir)
expert_policy_file = args['expert_policy_file']
expert_policy = load_policy(expert_policy_file)
N = args["dagger_N"]
num_rollouts = args["num_rollouts"]
max_timesteps = args["max_timesteps"]
render = args["render"]
returns = dagger(env, model, expert_policy, num_rollouts, max_timesteps, N,
render)
if args.get('results_file') is not None:
results = args.copy()
results["returns"] = returns
results["timestamp"] = datetime.now().isoformat()
dump_results(args['results_file'], results)
else:
print('returns', returns)
print('mean returns', map(np.mean, returns))
print('std of returns', map(np.std, returns))
def main():
import argparse
parser = argparse.ArgumentParser()
parser.add_argument('expert_policy_file', type=str)
parser.add_argument('envname', type=str)
parser.add_argument('--render', action='store_true')
parser.add_argument("--max_timesteps", type=int)
parser.add_argument('--num_rollouts',
type=int,
default=20,
help='Number of expert roll outs')
parser.add_argument('--results_file',
type=str,
help='File path for dumping the results')
args = parser.parse_args()
print('loading and building expert policy')
policy_fn = load_policy.load_policy(args.expert_policy_file)
print('loaded and built')
with tf.Session():
tf_util.initialize()
import gym
env = gym.make(args.envname)
max_steps = args.max_timesteps or env.spec.timestep_limit
returns = []
observations = []
actions = []
for i in range(args.num_rollouts):
print('iter', i)
obs = env.reset()
done = False
totalr = 0.
steps = 0
while not done:
action = policy_fn(obs[None, :])
observations.append(obs)
actions.append(action)
obs, r, done, _ = env.step(action)
totalr += r
steps += 1
if args.render:
env.render()
if steps % 100 == 0: print("%i/%i" % (steps, max_steps))
if steps >= max_steps:
break
returns.append(totalr)
print('returns', returns)
print('mean return', np.mean(returns))
print('std of return', np.std(returns))
if args.results_file is not None:
results = vars(args).copy()
results['returns'] = returns
results["timestamp"] = datetime.now().isoformat()
dump_results(args.results_file, results)
expert_data = {
'observations': np.array(observations),
'actions': np.array(actions)
}
dump_file = ("./expert_data/{}-{}.pkl"
"".format(args.envname, args.num_rollouts))
with open(dump_file, "wb") as f:
pickle.dump(expert_data, f)
# get all csv files in input directory
reg_x = re.compile(r'\.(csv)')
csv_input_files = []
for path, dnames, fnames in os.walk(csv_input_directory):
csv_input_files.extend(
[os.path.join(path, f) for f in fnames if reg_x.search(f)])
csv_input_files.sort()
modelFactoryInstance = modelFactory()
for m in models:
output_files = []
output_files.append("file,mse,parameters")
nan_output_files = []
nan_output_files.append("file")
helpers.dump_results(output_files, csv_output_directory, m)
print("##### [" + m + "]" + str(len(csv_input_files)) +
" CSV input files to process #####")
count = 1
for f in csv_input_files:
print("Processing [" + m + "]" + f)
if f.split("/")[-1] == "electricity_utilization_specific_user":
parameters["training_ratio"] = 0.58979536887
else:
parameters["training_ratio"] = defualt_training_ratio
# fetching input file data
dataframe_expect = pandas.read_csv(f)
value = np.array(dataframe_expect['value'])
timestamp = np.array(dataframe_expect['timestamp'])
model = "sherlock-framework-lstmcnn"
tesing_start = 362
list_file = "../results/"
model_results_folder = results_folder + "data/" + model + "/"
reg_x = re.compile(r'\.(csv)')
csv_input_files = []
for path, dnames, fnames in os.walk(csv_input_directory):
csv_input_files.extend([os.path.join(path, f) for f in fnames if reg_x.search(f)])
csv_input_files.sort()
output_files = []
output_files.append("file,mse,parameters")
helpers.dump_results(output_files, results_folder,model)
for f in csv_input_files:
try :
# fetching input file data
dataframe_expect = pandas.read_csv(f)
value = np.array(dataframe_expect['value'])
timestamp = np.array(dataframe_expect['timestamp'])
try :
label = np.array(dataframe_expect['label'])
except KeyError:
print "Warnning :["+f+"] No `label` colomn found data set without labels !. Assumes there are no anomolies"
label = np.zeros(len(value))
dataframe_predction = pandas.read_csv(helpers.get_result_file_name(f,results_folder,model))
prediction = np.array(dataframe_predction['prediction'])
params = "training_size="+str(tesing_start)