Data preparation with recipes was introduced as a beta feature in the Winter ‘17 release, but is now generally available and already turned on for you in Wave.
What Is a Dataset Recipe?
A dataset recipe is a saved set of transformations, or steps, to perform on a specific source dataset or replication. When you run a recipe, it applies the transformations and outputs the results to a new target dataset. You can schedule a recipe to run on a recurring basis to keep your target dataset up to date.
Use a recipe to combine data from different sources, and modify field values to ensure consistency in the new dataset you create. You can use the new dataset as a standalone dataset for exploration or dashboard creation, or in your dataflows or other recipes.
Create a Dataset Recipe
You create, run, schedule, and manage your recipes all in one place—the data manager.
Head to the Prepare tab of the data manager, where you can access your existing recipes and datasets.
Go to any dataset or replication in the data manager to create a recipe for it, or click Create Recipe to select your source data and get started. All your data preparation takes place on a single dataset recipe page, so all the tools you need are close at hand.
The recipe preview (1) gives you a real-time preview of your data as you prepare it. If you’re feeling overwhelmed by too many fields or rows, click the above the preview to change what you see. As you work on your data, your changes appear as recipe steps in the left pane (2). When you can’t find a field, use the field navigator in the right pane (3) to search for a field. If you see a field that you don’t need in your preview, click the to hide it. You can toggle the field navigator on and off with the button at the top of the page.
Prepare the Data
Your data preparation tools are all there for you at the top of the recipe page:
Let’s say you have a US Leads dataset that you are preparing for use in a marketing dashboard. Here’s what you can do.
Add fields from a lookup dataset or replication. Using a merge key field to match rows in your recipe with rows in your lookup, select the lookup fields you want to bring into your recipe.
For example, join a geodata dataset to your leads dataset using the zip code as a merge key.
Then select the geodata fields.
Filter the rows in your recipe.
For example, filter out leads that have a PO box zip code.
Add a field to bucket values from a specified dimension or measure field.
For example, add a field to categorize leads according to the number of people in the household.
Add a formula field to calculate new values from measures in your recipe using arithmetic operators and math functions.
For example, calculate how the household size for each lead compares to the national average.
Transform values in a specified dimension field. You can change case, split values using a delimiter, truncate values, or replace values with new ones.
For example, change category values to match the values that marketing uses.
Navigate Your Changes
As you prepare your data, each change you make appears as a recipe step on the left. Don’t worry if you don’t get it right the first time. You can revisit each step to make changes.
|Hover and click to see options.||Go to takes you back and forward to see your data at different steps.||Edit lets you make changes. Make changes to formula fields and transformations inline!|
Create the Dataset
You can save the recipe at any time to come back to it later. When the recipe is complete, click Create Dataset.