def train_autonas(ui):
train_nas = Trainer()
if ui.mode == 'train':
if ui.optimize == 'pso':
print("LOOKS GOOD - pso")
train_nas.pso_train(ui)
elif ui.optimize == 'aco':
print("LOOKS GOOD - aco ")
train_nas.aco_train(ui)
elif ui.optimize == 'autokeras' or ui.optimize == 'automodel':
train_nas.autokeras_train(ui)
if ui.mode == 'pretrain': # custom_model will be used within pre_train
featuresTrain, featuresVal, trainY, testY = train_nas.pre_train(ui)
if ui.second_model:
print("Running multiple classifiers")
train_nas.multiple_classifiers(featuresVal, testY, ui)
if ui.tune:
# tuning in this context means fine tuning of hyper parameters of chosen secondary model
# tuning of inner layers of pre-trained models also to be implemented
print("Tuning the secondary model")
train_nas.tune_secondary_model(featuresTrain, featuresVal, trainY,
testY, ui)
if ui.mode == 'hybrid': # custom_model will be used within pre_train
print("LOOKS GOOD - hybrid")
train_nas.hybrid(ui)
if ui.mode == 'ensemble':
print("LOOKS GOOD - ensemble ")
ensemble(ui)
if ui.mode == 'super_stacker':
print("LOOKS GOOD - super_stacker ")
super_stacker(ui)
return
import matplotlib.pyplot as plt
from variables import multi_step_lags, test_len, full_len
import os
root_path = os.path.dirname(os.path.abspath('__file__'))
import sys
sys.path.append(root_path)
from ensembler import ensemble
# Set the project parameters
ORIGINAL = 'XianyangRunoff1951-2018(1953-2018).xlsx'
STATION = 'Xianyang'
DECOMPOSER = 'ssa'
PREDICTOR = 'esvr' # esvr or gbrt or lstm
PREDICT_PATTERN = 'multi_step_1_month_forecast' # hindcast or forecast
ensemble(
root_path=root_path,
original_series=ORIGINAL,
station=STATION,
decomposer=DECOMPOSER,
multi_step_llags_dict=variables['lags_dict'],
predictor=PREDICTOR,
predict_pattern=PREDICT_PATTERN,
test_len=test_len,
full_len=full_len,
)
plt.show()