Skip to content

7. Create Observation Tables

An Observation Table is a structured collection of historical data points that acts as the foundation for training datasets. By adding features, you can create Feature Tables that can be used to train and validate Machine Learning models.

Each data point represents a specific historical moment for a particular entity and may also include target values. Observation Tables are often utilized across experiments within the same use case, even if selected features and models vary.

This guide explains how to upload and configure Observation Tables in FeatureByte and link them to your context and use cases.

Step 1: Navigate to Observation Table Catalog

From the menu, navigate to the 'Formulate' section and select the Observation Table catalog.

Empty Observation Table Catalog

Step 2: Upload Observation Table

  1. Click Image
  2. Select 'Upload file' tab and configure the table:
    • Name: "In_Store_Customer_2023_10K"
    • Purpose: "EDA" as we will use the table to run EDA analysis
    • Primary Entity: "customer"
    • Target: Leave blank
    • Context: Leave blank
    • CSV/Parquet: "In-Store Customer_2023_10K_sample.parquet" that you can download here.

Name

  1. Select the uploaded table from the catalog.
  2. Go to the 'About' tab and choose "In-Store Customer" from the Context dropdown menu.

Name

Step 4: Create Observation Table with Target Values

Option A: Using the SDK Target

  1. Ensure you have created the target "CUSTOMER_More_than_1_purchase_next_7d" using the FeatureByte SDK.
  2. Open the 'About' tab, scroll down and click Compute New Observation Table With Target
  3. Select the target and new the new table, e.g., "Pre_Purchase_Customer_Activity_next_week_2023_10K".

Name

Option B: Upload Observation Table with Target Values

If you don't have access to the FeatureByte SDK, follow these steps.

  1. Click Image
  2. Upload an Observation Table with pre-populated target values:
    • Name: "Pre_Purchase_Customer_Activity_next_week_2023_10K_M"
    • Purpose: "EDA" as we will use the table to run EDA analysis
    • Primary Entity: "customer"
    • Target: "CUSTOMER_More_than_1_purchase_next_7d_M"
    • CSV/Parquet: "Pre_Purchase_Customer_Activity_next_week_2023_M.parquet" that you can download here.

Name

Step 5: Check Observation Tables

Confirm successful registration by reviewing the Observation Table Catalog.

Name

Step 6: Preview Observation Table

  1. Locate your observation table.
  2. Go to the 'Preview' tab and ensure the target column appears.

Name

Step 7: Set Observation Table as the Default EDA Table for the Use Case

The Default EDA Table is used for feature analysis in the use case.

If you made the target using the SDK, follow these steps to set the Default EDA Table.

  1. Navigate to the Use Case Catalog and select "Customer Activity Next Week before a purchase".

  2. Set "Pre_Purchase_Customer_Activity_next_week_2023_10K" as the Default EDA Table.

Name

Name

Ensure, once completed, your setup resembles the image below.

Name

If you uploaded an Observation Table with pre-populated target values, follow these steps to set the Default EDA Table.

  1. Select the uploaded table "Pre_Purchase_Customer_Activity_next_week_2023_10K_M" and link it to the "In-Store Customer" Context.
  2. Navigate to the Use Case Catalog and select "Customer Activity Next Week before a purchase (using the descriptive target)".
  3. Set "Pre_Purchase_Customer_Activity_next_week_2023_10K_M" as the Default EDA Table.

Name

Name

Step 8: Verify the setting by checking the Observation Table Catalog.

Name