def main(args):
assert pyro.__version__ >= "0.4.1"
pyro.enable_validation(__debug__)
pyro.set_rng_seed(args.seed)
dataset = load_hourly_od(args)
if args.tiny:
dataset["stations"] = dataset["stations"][:args.tiny]
dataset["counts"] = dataset["counts"][:, :args.tiny, :args.tiny]
forecaster = train(args, dataset)
if forecaster is None:
return
window_begin = max(0, args.truncate - args.batch_size)
window_end = args.truncate
truth = dataset['counts'][window_end:window_end + args.forecast_hours]
forecast = forecaster(window_begin,
window_end,
args.forecast_hours,
num_samples=args.num_samples)
assert forecast.shape == (args.num_samples, ) + truth.shape
log_prob = forecaster.log_prob(window_begin, window_end,
truth) / truth.numel()
torch.save({
'forecast': forecast,
'truth': truth,
'log_prob': log_prob,
}, args.forecast_filename)
def main(args):
assert pyro.__version__ >= "0.4.1"
pyro.enable_validation(__debug__)
pyro.set_rng_seed(args.seed)
dataset = load_hourly_od(args)
forecaster = train(args, dataset)
num_samples = 10
forecast = forecaster(0, 24 * 7, 24)
assert forecast.shape == (24, ) + dataset["counts"].shape[-2:]
forecast = forecaster(0, 24 * 7, 24, num_samples=num_samples)
assert forecast.shape == (num_samples, 24) + dataset["counts"].shape[-2:]
return forecast
def main(argv):
config = ForecastConfig()
loadConfig = ForecastLoadConfig()
if not loadConfig.LOAD_MODEL:
assert not os.path.isdir(loadConfig.OUTPUT_FOLDER)
os.makedirs(loadConfig.OUTPUT_FOLDER)
train_part, validation_part, test_part, scaler = getNormalizedParts(
config, loadConfig, config.TIMESTAMPS)
train_x, train_y = buildSet(train_part, loadConfig.LOOK_BACK,
loadConfig.OUTPUT_SIZE)
validation_x, validation_y = buildSet(validation_part,
loadConfig.LOOK_BACK,
loadConfig.OUTPUT_SIZE)
test_x, test_y = buildSet(test_part, loadConfig.LOOK_BACK,
loadConfig.OUTPUT_SIZE)
end_train, end_validation = get_split_indexes(config)
if not loadConfig.LOAD_MODEL:
copyfile("./code/forecast_conf.py",
loadConfig.OUTPUT_FOLDER + "forecast_conf.py")
copyfile(
"./code/forecast_load_conf.py",
loadConfig.OUTPUT_FOLDER + "forecast_load_conf.py",
)
model = buildModel(loadConfig, train_x.shape)
history = train(loadConfig, model, train_x, train_y, validation_x,
validation_y)
saveModel(loadConfig, model)
plotHistory(loadConfig, history)
validation_timestamps = config.TIMESTAMPS[end_train:end_validation]
validation_y_timestamps = validation_timestamps[loadConfig.LOOK_BACK:]
assert (len(validation_y_timestamps) == len(validation_y) +
loadConfig.OUTPUT_SIZE)
validation_prediction = model.predict(validation_x)
validation_mse = mean_squared_error(validation_y,
validation_prediction)
print("validation mse: ", validation_mse)
plotPredictionPart(
loadConfig,
validation_y[1, :],
validation_prediction[1, :],
"1st day of validation set",
validation_y_timestamps[:loadConfig.OUTPUT_SIZE],
"paramtuning1",
)
else:
model = loadModel(loadConfig)
test_prediction = model.predict(test_x)
test_mse = mean_squared_error(test_y, test_prediction)
print("test mse: ", test_mse)
test_timestamps = config.TIMESTAMPS[end_validation:end_validation +
len(test_part)]
test_y_timestamps = test_timestamps[loadConfig.LOOK_BACK:]
assert len(test_y_timestamps) == len(test_y) + loadConfig.OUTPUT_SIZE
plot_multiple_days(config, loadConfig, test_part[:, 0],
test_prediction, test_timestamps)
plotPrediction(
train_y,
model.predict(train_x),
validation_y,
model.predict(validation_x),
test_y,
test_prediction,
config.TIMESTAMPS,
loadConfig,
)
plotPredictionPart(
loadConfig,
test_y[1, :],
test_prediction[1, :],
"1st day of test set",
test_y_timestamps[:loadConfig.OUTPUT_SIZE],
"paramtuning2",
)
def buildModelPv(trainX, trainY, valX, valY, config_pv):
model = buildModel(config_pv, trainX.shape)
# training it
history = train(config_pv, model, trainX, trainY, valX, valY)
saveModel(config_pv, model)
return model, history