Web cut#
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
plt.rcParams['figure.dpi'] = 150
doe = pd.read_csv('../data/doe.csv')
data = pd.read_csv('../data/cut_web_all.csv')
data.drop(data[data.doe_id == 1000].index, inplace=True)
data.drop(data[data.doe_id == 247].index, inplace=True)
from mesh_predictor import CutPredictor
reg = CutPredictor()
reg.load_data(
doe = doe,
data = data,
index='doe_id',
process_parameters = [
'Blechdicke',
'Niederhalterkraft',
'Ziehspalt',
'Einlegeposition',
'Ziehtiefe',
'Rp0',
],
categorical = [
'Ziehspalt',
'Ziehtiefe',
],
position = 'c_phi',
output = 'c_rho',
validation_split=0.1,
validation_method='leaveoneout'
)
reg.save_config("../models/cut_web.pkl")
reg.data_summary()
best_config = reg.autotune(
save_path='../models/best_web_model',
trials=100,
max_epochs=100,
layers=[2, 4],
neurons=[64, 256, 64],
dropout=[0.0, 0.5, 0.1],
learning_rate=[1e-5, 1e-3]
)
print(best_config)
config = {
'batch_size': 4096,
'max_epochs': 100,
'layers': [128, 128, 128, 128, 128],
'dropout': 0.0,
'learning_rate': 0.01
}
# or best_config from autotune if you already did it once
reg.custom_model(save_path='../models/best_web_model', config=config, verbose=True)
reg.training_summary()
idx = np.random.choice(1000) + 1
print("Doe_ID", idx)
reg.compare(idx)
%matplotlib inline
plt.rcParams['figure.dpi'] = 150
def viz(x, y):
plt.figure()
plt.plot(x, y[:, 0])
plt.xlabel('c_phi')
plt.ylabel('c_rho')
reg.interactive(function=viz, positions=100)