vertex_ai_tabular_dataset_creation_from_gcs_csv.py

1from google.cloud import aiplatform
2
3def create_dataset_tabular_sample(
4    project: str,
5    location: str,
6    display_name: str,
7    gcs_source: str,
8):
9    """
10    Create a tabular dataset and import data from Google Cloud Storage.
11    Args:
12        project: The project ID.
13        location: The region (e.g., us-central1).
14        display_name: The name of the dataset.
15        gcs_source: The Cloud Storage URI of the CSV file.
16    """
17    # Initialize the Vertex AI SDK
18    aiplatform.init(project=project, location=location)
19
20    # Create the dataset
21    dataset = aiplatform.TabularDataset.create(
22        display_name=display_name,
23        gcs_source=gcs_source,
24    )
25
26    dataset.wait()
27
28    print(f"Dataset created. Resource name: {dataset.resource_name}")
29    return dataset
30
31# Example usage (placeholders):
32# create_dataset_tabular_sample(
33#     project="your-project-id",
34#     location="us-central1",
35#     display_name="my_tabular_dataset",
36#     gcs_source="gs://cloud-samples-data/ai-platform- deception/census_data.csv"
37# )