def test_model_after_train(calc_train,
real_vals_prop_train,
calc_test,
real_vals_prop_test,
directory,
prop_name='logP'):
"""Scatter plot comparing ground truth data with the modelled data";
includes both test and training data."""
plt.figure()
plt.scatter(calc_train,
real_vals_prop_train,
color='red',
s=40,
facecolors='none')
plt.scatter(calc_test,
real_vals_prop_test,
color='blue',
s=40,
facecolors='none')
plt.xlim(min(real_vals_prop_train) - 0.5, max(real_vals_prop_train) + 0.5)
plt.ylim(min(real_vals_prop_train) - 0.5, max(real_vals_prop_train) + 0.5)
plt.xlabel('Modelled ' + prop_name)
plt.ylabel('Computed ' + prop_name)
plt.title('Train set (red), test set (blue)')
name = directory + '/test_model_after_training'
plt.savefig(name)
closefig()
def dreamed_histogram(prop_lst, prop, directory, prop_name='logP'):
"""Plot distribution of property values from a given list of values
(after transformation)"""
plt.figure()
plt.hist(prop_lst, density=True, bins=30)
plt.ylabel(prop_name + ' - around ' + str(prop))
name = directory + '/dreamed_histogram'
plt.savefig(name)
closefig()
def running_avg_test_loss(avg_test_loss, directory):
"""Plot running average test loss"""
plt.figure()
plt.plot(avg_test_loss)
plt.xlabel('Epochs')
plt.ylabel('Running average test loss')
name = name = directory + '/runningavg_testloss'
plt.savefig(name)
closefig()
def prediction_loss(train_loss, test_loss, directory):
"""Plot prediction loss during training of model"""
plt.figure()
plt.plot(train_loss, color='red')
plt.plot(test_loss, color='blue')
plt.title('Prediction loss: training (red), test (blue)')
plt.xlabel('Epochs')
plt.ylabel('Loss')
name = directory + '/predictionloss_test&train'
plt.savefig(name)
closefig()
def initial_histogram(prop_dream,
directory,
dataset_name='QM9',
prop_name='logP'):
"""Plot distribution of property values from a given list of values
(before transformation)"""
plt.figure()
plt.hist(prop_dream, density=True, bins=30)
plt.ylabel(prop_name + ' - ' + dataset_name)
name = directory + '/QM9_histogram'
plt.savefig(name)
closefig()
def test_model_before_dream(trained_data_prop,
computed_data_prop,
directory,
prop_name='logP'):
"""Scatter plot comparing ground truth data with modelled data"""
plt.figure()
plt.scatter(trained_data_prop, computed_data_prop)
plt.xlabel('Modelled ' + prop_name)
plt.ylabel('Computed ' + prop_name)
name = directory + '/test_model_before_dreaming'
plt.savefig(name)
plt.show()
closefig()