def res():
comp = request.args.get('comp', None)
budget = request.args.get('budget', None)
print(comp, budget)
# Searches for mega pc
mega = search_megapc(comp, budget)
mega_prd = mega['prd'].encode('ascii', 'ignore').decode('unicode_escape')
# Searches for sbs
sbs = search_sbs(comp, budget)
sbs_prd = sbs['prd'].encode('ascii', 'ignore').decode('unicode_escape')
# Search extreme
extrme = search_extreme(comp, budget)
extrme_prd = extrme['prd'].encode('ascii',
'ignore').decode('unicode_escape')
# Search tunisianet
tn = search_tunisia(comp, budget)
tn_prd = tn['prd'].encode('ascii', 'ignore').decode('unicode_escape')
win1 = benchmark(sbs, tn, comp)
win2 = benchmark(mega, extrme, comp)
final = benchmark(win1, win2, comp)
result = {'data': [mega, sbs, extrme, tn], 'win': final}
return jsonify(result)
def main():
print "ZCML took %.3f seconds." % measure(load_ftesting_zcml)
print "Setup took %.3f seconds." % measure(setup_benchmark)
benchmark("Daily calendar view with many recurrent events", daily_view)
benchmark("Weekly calendar view with many recurrent events", weekly_view)
benchmark("Monthly calendar view with many recurrent events", monthly_view)
benchmark("Yearly calendar view with many recurrent events", yearly_view)
def main():
print "ZCML took %.3f seconds." % measure(load_ftesting_zcml)
print "Setup took %.3f seconds." % measure(setup_benchmark)
benchmark("Daily calendar view with many simple events", daily_view)
benchmark("Weekly calendar view with many simple events", weekly_view)
benchmark("Monthly calendar view with many simple events", monthly_view)
benchmark("Yearly calendar view with many simple events", yearly_view)
def train(training_data, model_path=DEFAULT_DATA_PATH, test_data=None):
X_train, y_train = read_train_data(training_data)
logging.info("Training on {} examples for {} labels".format(len(X_train), len(set(y_train))))
logging.info("Starting the training")
prediction_pipeline.fit(preprocess_pipeline.fit_transform(X_train, y_train), y_train)
if test_data != None:
X_test, y_test = read_test_data(test_data, y_train)
logging.info("Evaluating the model")
X_test = preprocess_pipeline.transform(X_test)
benchmark(prediction_pipeline, X_train, y_train, X_test, y_test, verbose=2)
logging.info("Storing the model to {}".format(model_path))
joblib.dump(prediction_pipeline, model_path)
def main():
print "ZCML took %.3f seconds." % measure(load_ftesting_zcml)
print "Setup took %.3f seconds." % measure(setup_benchmark)
benchmark("Daily calendar view on the start date.", daily_view_start_date)
benchmark("Daily calendar view a year after the start date.",
daily_view_in_a_year)
benchmark("Daily calendar view ten years after the start date.",
daily_view_in_ten_years)
def main():
print "ZCML took %.3f seconds." % measure(load_ftesting_zcml)
print "Setup took %.3f seconds." % measure(setup_benchmark)
benchmark("Daily calendar view on the start date.",
daily_view_start_date)
benchmark("Daily calendar view a year after the start date.",
daily_view_in_a_year)
benchmark("Daily calendar view ten years after the start date.",
daily_view_in_ten_years)
def run_benchmark(function, label):
config = BenchmarkConfig(
batch_sizes=[2**i for i in range(FLAGS.batches)],
device=FLAGS.device,
dt=FLAGS.dt,
label=label,
runs=FLAGS.runs,
sequence_length=FLAGS.sequence_length,
start=FLAGS.start,
stop=FLAGS.stop,
step=FLAGS.step,
)
collector = partial(collect, label=label)
results = benchmark(function, collector, config)
timestamp = time.strftime("%Y-%m-%d-%H-%M-%S")
filename = f"{timestamp}-{label}.csv"
pd.DataFrame(results).to_csv(filename)
'autosklearn-v': AutoSklearnVanillaBenchmark,
'autosklearn-m': AutoSklearnMetaBenchmark,
'autosklearn-e': AutoSklearnEnsBenchmark,
'tpot': TPOTBenchmark,
'recipe': RecipeBenchmark
}
if model in model_to_bench:
model_to_bench[model]().benchmark(dataset_file,
output_file,
time_limit=time,
dataset_test_file=dataset_test_file,
config=config)
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('input_file', help='Dataset file')
parser.add_argument('output_file', help='Benchmark result file')
parser.add_argument('-t', '--time', type=int, help='Time budget')
parser.add_argument('-m', '--model', help='AutoML Model')
parser.add_argument('-te', '--test_file', help='Dataset test file')
parser.add_argument('-c', '--config', nargs='*')
args = parser.parse_args()
benchmark(dataset_file=args.input_file,
output_file=args.output_file,
time=args.time,
model=args.model,
dataset_test_file=args.test_file,
config=args.config)