imputer_text.fit(train_df=df_train,
learning_rate=1e-4,
num_epochs=50,
final_fc_hidden_units=[512])
#Fit a Model With HPO
imputer_text = SimpleImputer(
input_columns=['title', 'text'],
output_column='finish',
output_path='imputer_model',
)
imputer_text.fit_hpo(train_df=df_train,
num_epochs=50,
learning_rate_candidates=[1e-3, 1e-4],
final_fc_hidden_units_candidates=[[100]],
num_hash_bucket_candidates=[2**10, 2**15],
tokens_candidates=['chars', 'words'])
#------------------------------------------------------------------------------------
'''
Numerical Data
'''
#Generate synthetic numerical data
numeric_data = np.random.uniform(-np.pi, np.pi, (n_samples, ))
df = pd.DataFrame({
'x':
numeric_data,
'*2':
numeric_data * 2. + np.random.normal(0, .1, (n_samples, )),
'**2':
# Print overall classification report
print(
classification_report(predictions['finish'],
predictions['finish_imputed']))
# ------------------------------------------------------------------------------------
"""
Run SimpleImputer with hyperparameter optimization
"""
# Initialize a SimpleImputer model
imputer = SimpleImputer(input_columns=['title', 'text'],
output_column='finish',
output_path='imputer_model')
# Fit an imputer model with default list of hyperparameters
imputer.fit_hpo(train_df=df_train)
# Fit an imputer model with customized HPO
imputer.fit_hpo(train_df=df_train,
num_epochs=5,
patience=3,
learning_rate_candidates=[1e-3, 1e-4],
num_hash_bucket_candidates=[2**15],
tokens_candidates=['words', 'chars'])
# ------------------------------------------------------------------------------------
"""
Load saved model and get metrics from SimpleImputer
"""
# Load saved model
imputer = SimpleImputer.load('./imputer_model')
tokens='chars')
imputer_text.fit(train_df=df_train,
learning_rate=1e-4,
num_epochs=5,
final_fc_hidden_units=[512])
# Fit a Model With HPO
imputer_text = SimpleImputer(
input_columns=['title', 'text'],
output_column='finish',
output_path='imputer_model',
)
imputer_text.fit_hpo(train_df=df_train,
num_epochs=5,
num_hash_bucket_candidates=[2**10, 2**15],
tokens_candidates=['chars', 'words'])
# ------------------------------------------------------------------------------------
"""
Numerical Data
"""
# Generate synthetic numerical data
n_samples = 100
numeric_data = np.random.uniform(-np.pi, np.pi, (n_samples, ))
df = pd.DataFrame({
'x':
numeric_data,
'*2':
numeric_data * 2. + np.random.normal(0, .1, (n_samples, ))
})