Using Training Examples

Contents

  1. What are training examples?

  2. Common Use Cases

  3. Getting Started

  4. Providing Multiple Training Examples

  5. Color Coding

  6. Editing or Deleting Training Examples

  7. Providing an Empty or Null Training Example

What are training examples?

Training examples are used to quickly transform and map your source data to the columns in the output table. It involves providing examples of the desired output to teach our AI to detect a pattern and create a program that auto-populates the remaining cells for that column with the transformed data. You can provide several training examples for a given column to increase the complexity of the program learned by our AI.

Common Use Cases

Some common use cases involve combining last name and first name from two source columns, extracting a zip code from a full address, or changing the format of a date.

Getting Started

Step 1: Click on a cell in the output table to get started.

Step 2: Look within the source table (on the left) to find the column(s) that contain(s) the data you want to extract and map to the selected column in the output table. Go to the row that is highlighted green, and copy the value(s) from the input column(s) and paste it into the selected cell.

Example: In the example below, we want to extract the first name from the full name column. We have copied "Adam" from the first row and pasted it into the first cell of the FirstName output column.

Step 3: After pasting the value into the cell, press enter or click on the blue enter button to submit the training example.

The AI program will output values for each cell based on the first training example. Sometimes the transformation is simple enough to output the correct value for each cell, but you may need to provide additional training examples to refine the program.

Example: In the example below, you can see that we extracted the first name for each row based on the one training example.

Step 4: Review the output for each cell in that column and confirm that there are no errors. Look for cells that are highlighted red or orange and provide another training example where required.

Providing Multiple Training Examples

One training example may be enough for a simple transformation, but sometimes a few training examples will be required for our AI to detect a pattern and transform each row correctly.

After providing the first training example, review the output to confirm if you need to provide additional training examples.

Color Coding

Cells with Black Borders

Cells that contain training examples or formulas have black borders. If you want to adjust the output for that column, click on these cells to delete or edit them.

Cells with Orange Borders

Cells that have been flagged for review have orange borders. Our AI automatically detects potential values that are incorrect. Review these rows and provide another training example if they are incorrect.

Cells with Red Borders

Cells that require an additional training example have red borders. This occurs when our AI does not know how to transform the data for that row based on the already provided training examples. Click on these cells and provide another training example.

Editing or Deleting Training Examples

If you want to edit a training example, click on the cell that contains the training example, edit the training example, then press enter to submit the revised example.

If you want delete a specific training example, click on the cell that contains the training example, then click on the delete button to the right of that cell.

Providing an Empty or Null Training Example

If you want to provide an empty or null training example for a row that does not have a source value, click on the output cell, leave the cell blank, then press enter.

If the column is nullable field, we will insert a (null) value, as show below.

If the column is a required field, we will insert an (empty string), as shown below.

‚Äč