Add a check for data that is exclusively numerical

[franchuterivera]

There seem to be some failures if the data is exclusively numerical in the preprocessing part.

To reproduce, switch the tabular example to a numerical only dataset (mnist maybe?) and the fitting will fail.

[ravinkohli]

Hey, I could not reproduce the error on my machine. I tried the example with mnist and everything worked

[franchuterivera]

Here you go:

"""
======================
Tabular Classification
======================
"""
import numpy as np

import sklearn.datasets
import sklearn.model_selection
from sklearn.utils.multiclass import type_of_target

from autoPyTorch.pipeline.tabular_classification import TabularClassificationPipeline

# Get the training data for tabular classification
# Move to Australian to showcase numerical vs categorical
X, y = sklearn.datasets.fetch_openml(data_id=40996, return_X_y=True, as_frame=True)
X_train, X_test, y_train, y_test = sklearn.model_selection.train_test_split(
    X,
    y,
    random_state=1,
    test_size=0.2,
)

train_indices, val_indices = sklearn.model_selection.train_test_split(
    list(range(X_train.shape[0])),
    random_state=1,
    test_size=0.25,
)

output_type = type_of_target(y)
print(f"X_train={X_train.shape} train_indices={train_indices} output_type={output_type}")

# Mock the categories
categorical_columns = []
numerical_columns = X.columns
categories = [np.unique(X[a]).tolist() for a in categorical_columns]

# Create a proof of concept pipeline!
dataset_properties = {
    'task_type': 'tabular_classification',
    'categorical_columns': categorical_columns,
    'numerical_columns': numerical_columns,
    'output_type': output_type,
}
pipeline = TabularClassificationPipeline(dataset_properties=dataset_properties)

# Configuration space
pipeline_cs = pipeline.get_hyperparameter_search_space()
print("Pipeline CS:\n", '_' * 40, f"\n{pipeline_cs}")
config = pipeline_cs.sample_configuration()
print("Pipeline Random Config:\n", '_' * 40, f"\n{config}")
pipeline.set_hyperparameters(config)

# Fit the pipeline
print("Fitting the pipeline...")
pipeline.fit(X={
    'categorical_columns': categorical_columns,
    'numerical_columns': numerical_columns,
    'num_features': X.shape[1],
    'num_classes': len(np.unique(y)),
    'is_small_preprocess': True,
    'categories': categories,
    'X_train': X_train,
    'y_train': y_train,
    'train_indices': train_indices,
    'val_indices': val_indices,
    'X_test': X_test,
    'y_test': y_test,
    'dataset_properties': dataset_properties,
    # Training configuration
    'job_id': 'example_tabular_classification_1',
    'device': 'cpu',
    'budget_type': 'epochs',
    'epochs': 100,
    'torch_num_threads': 1,
    'early_stopping': 20,
})

# Showcase some components of the pipeline
print(pipeline)

# Showcase performance of pipeline
print(pipeline.named_steps['trainer'].run_summary.repr_last_epoch())

[ravinkohli]

Tried it, still could not reproduce the error. It worked fine. Could you tell me what the error was?

[franchuterivera]

Here is the error for help:

  File "debug.py", line 46, in <module>
    pipeline = TabularClassificationPipeline(dataset_properties=dataset_properties)
  File "/home/chico/work/temporal/Auto-PyTorch_refactor/autoPyTorch/pipeline/tabular_classification.py", line 66, in __init__
    super().__init__(
  File "/home/chico/work/temporal/Auto-PyTorch_refactor/autoPyTorch/pipeline/base_pipeline.py", line 74, in __init__
    self.config_space = self.get_hyperparameter_search_space()
  File "/home/chico/work/temporal/Auto-PyTorch_refactor/autoPyTorch/pipeline/base_pipeline.py", line 191, in get_hyperparameter_search_space
    self.config_space = self._get_hyperparameter_search_space(
  File "/home/chico/work/temporal/Auto-PyTorch_refactor/autoPyTorch/pipeline/tabular_classification.py", line 176, in _get_hyperparameter_search_space
    cs = self._get_base_search_space(
  File "/home/chico/work/temporal/Auto-PyTorch_refactor/autoPyTorch/pipeline/base_pipeline.py", line 312, in _get_base_search_space
    sub_config_space = node.get_hyperparameter_search_space(
  File "/home/chico/work/temporal/Auto-PyTorch_refactor/autoPyTorch/pipeline/components/preprocessing/tabular_preprocessing/scaling/base_scaler_choice.py", line 75, in get_hyperparameter_search_space
    if not dataset_properties['numerical_columns']:
  File "/home/chico/anaconda3/envs/autopytorch/lib/python3.8/site-packages/pandas-1.1.3-py3.8-linux-x86_64.egg/pandas/core/indexes/base.py", line 2419, in __nonzero__
    raise ValueError(
ValueError: The truth value of a Index is ambiguous. Use a.empty, a.bool(), a.item(), a.any() or a.all().