def algo_run(model, pre_processing_pipeline, scaler, sampler, seed):
start_time = datetime.datetime.now()
df = utils.get_dataset()
df = pre_processing_pipeline[1](df)
X, y = utils.X_y_split(df)
if "Keras" in model[0]:
model_eval = model[1][:-1] + ",input_dim=" + str(X.shape[1]) + ")"
else:
model_eval = model[1]
model_eval = eval(model_eval)
try:
y_predicted = utils.cross_validation_average_results(
model_eval,
X,
y,
n_splits=5,
scaler=scaler[1],
sampling_technique=sampler[1])
threshold = utils.max_threshold(y_predicted,
y,
threshold_range=(0.2, 0.6),
iterations=1000,
visualization=False)
y_pred = utils.predict_with_threshold(y_predicted, threshold)
result = utils.profit_share(y_pred, y)
precision = utils.calculate_precision_score(y_pred, y)
recall = utils.calculate_recall_score(y_pred, y)
except:
result = -1
recall = -1
precision = -1
time_elapsed = datetime.datetime.now() - start_time
# Create result string
result_string = ",".join([
str(seed), model[0], pre_processing_pipeline[0], scaler[0], sampler[0],
str(time_elapsed),
str(precision),
str(recall),
str(result)
])
# Write result to a file
with open(file_name, "a") as myfile:
myfile.write(result_string + "\n")
# Output result to terminal
print(model[0] + ": " + str(result))
if result > 0.6:
print(
"!!!!!!!!!!!!!!!!!!!!!!!!!yey!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!1"
)
def algo_run(model, pre_processing_pipeline, scaler, sampler, seed):
start_time = datetime.datetime.now()
df = utils.get_dataset()
if pre_processing_pipeline[1] != None:
df = pre_processing_pipeline[1](df)
X, y = utils.X_y_split(df)
if "Keras" in model[0]:
model_eval = model[1][:-1] + ",input_dim=" + str(X.shape[1]) + ")"
elif "GP" in model[0]:
model_eval = model[1][:-1] + ",random_state=" + str(
seed) + ",feature_names=" + str(list(X.columns)) + ")"
elif "Tree" in model[0]:
model_eval = model[1][:-1] + ",random_state=" + str(seed) + ")"
elif "XG" in model[0]:
model_eval = model[1][:-1] + ",seed=" + str(seed) + ")"
else:
model_eval = model[1]
model_eval = eval(model_eval)
y_predicted = utils.cross_validation_average_results(
model_eval,
X,
y,
n_splits=5,
scaler=scaler[1],
sampling_technique=sampler[1])
mean_s_error = utils.calculate_mean_absolute_error(y_predicted, y)
explained_variance = utils.calculate_explained_variance_score(
y_predicted, y)
time_elapsed = datetime.datetime.now() - start_time
# Create result string
result_string = ",".join([
str(seed), model[0], pre_processing_pipeline[0], scaler[0], sampler[0],
str(time_elapsed),
str(mean_s_error),
str(explained_variance)
])
# Write result to a file
with open(file_name, "a") as myfile:
myfile.write(result_string + "\n")
print(model[0] + ": " + str(mean_s_error))
if 'GS_GP' in model[1]:
idx = model_eval._program.parents['parent_idx']
fade_nodes = model_eval._program.parents['parent_nodes']
print(model_eval._programs[-2][idx])
def algo_run(seed, model):
start_time = datetime.datetime.now()
df = utils.get_dataset()
X, y = utils.X_y_split(df)
model_eval = add_seed(model, seed, X)
model_eval = eval(model_eval)
X_train, X_test, y_train, y_test = train_test_split(X,
y,
test_size=0.20,
random_state=seed)
model_eval.fit(X_train, y_train)
y_predicted = model_eval.predict(X_test)
mean_s_error = utils.calculate_mean_absolute_error(y_predicted, y_test)
explained_variance = utils.calculate_explained_variance_score(
y_predicted, y_test)
time_elapsed = datetime.datetime.now() - start_time
# Create result string
log_parameters = [
seed, model[0], time_elapsed, mean_s_error, explained_variance
]
if 'GS_GP' in model[0]:
#idx = model_eval._program.parents['parent_idx']
if "p_gs_crossover=0.0" in model[0]:
fade_nodes = model_eval._program.parents['parent_nodes']
#log_parameters.append(len(model_eval._programs[-2][idx].program))
result_string = ",".join([str(value) for value in log_parameters])
# Write result to a file
with open(file_name, "a") as myfile:
myfile.write(result_string + "\n")
print(result_string)