Refining data

Prerequisites

Before you can refine data, you need to create a project that uses Cloud Object Storage.

If you have data in cloud or on-premises data sources, you'll need to add connections to those sources and you'll need to add data assets from each connection. If you want to be able to save refined data to cloud or on-premises data sources, create connections for this purpose as well. Source connections can be used only to read data; target connections can be used only to load (save) data. When you create a target connection, be sure to use credentials that have Write permission or you won't be able to save your Data Refinery flow output to the target.

Source file limitations

Target file limitations

The following limitation applies if you save Data Refinery flow output (the target data set) to a file:

• You can't change the file format if the file is an existing data asset.

Data protection rules

Data Refinery does not support data protection rules for row filtering. Data Refinery jobs might fail if the asset is governed by row filtering data protection rules. Additionally, if you add an asset from IBM Knowledge Catalog to a project that is governed by row-filtering data protection rules, the masking will not be enforced in Data Refinery. For information, see Data protection rules enforcement.

Data set previews

Data Refinery provides support for large data sets, which can be time-consuming and unwieldy to refine. To enable you to work quickly and efficiently, it operates on a subset of rows in the data set while you interactively refine the data. When you run a job for the Data Refinery flow, it operates on the entire data set.

Refine your data

The following video shows you how to refine data.

This video provides a visual method to learn the concepts and tasks in this documentation.

• Transcript
  Synchronize transcript with video

  Video transcript

  Time	Transcript
  00:00	This video shows you how to shape raw data using Data Refinery.
  00:05	To get started refining data from a project, view the data asset and open it in Data Refinery.
  00:14	The "Information" pane contains the name for the data flow and for the data flow output, once you've finished refining the data.
  00:23	The "Data" tab shows you a sample set of the rows and columns in the data set.
  00:29	To improve performance, you won't see all the rows in the shaper.
  00:33	But rest assured that when you are done refining the data, the data flow will be run on the full data set.
  00:41	The "Profile" tab shows you frequency and summary statistics for each of your columns.
  00:49	The "Visualizations" tab provides data visualizations for the columns you are interested in.
  00:57	Suggested charts have a blue dot next to their icons.
  01:03	Use the different perspectives available in the charts to identify patterns, connections, and relationships within the data.
  01:12	Now, let's do some data wrangling.
  01:17	Start with a simple operation, like sorting on the specified column - in this case, the "Year" column.
  01:27	Say you want to focus on delays just for a specific airline so you can filter the data to show only those rows where the unique carrier is "United Airlines".
  01:47	It would be helpful to see the total delay.
  01:50	You can do that by creating a new column to combine the arrival and departure delays.
  01:56	Notice that the column type is inferred to be integer.
  02:00	Select the departure delay column and use the "Calculate" operation.
  02:09	In this case, you'll add the arrive delay column to the selected column and create a new column, called "TotalDelay".
  02:23	You can position the new column at the end of the list of columns or next to the original column.
  02:31	When you apply the operation, the new column displays next to the departure delay column.
  02:38	If you make a mistake, or just decide to make a change, just access the "Steps" panel and delete that step.
  02:46	This will undo that particular operation.
  02:50	You can also use the redo and undo buttons.
  02:56	Next, you'd like to focus on the "TotalDelay" column so you can use the "select" operation to move the column to the beginning.
  03:09	This command arranges the "TotalDelay" column as the first in the list, and everything else comes after that.
  03:21	Next, use the "group_by" operation to divide the data into groups by year, month, and day.
  03:32	So, when you select the "TotalDelay" column, you'll see the "Year", "Month", "DayofMonth", and "TotalDelay" columns.
  03:44	Lastly, you want to find the mean of the "TotalDelay" column.
  03:48	When you expand the "Operations" menu, in the "Organize" section, you'll find the "Aggregate" operation, which includes the "Mean" function.
  04:08	Now you have a new column, called "AverageDelay", that represents the average for the total delay.
  04:17	Now to run the data flow and save and create the job.
  04:24	Provide a name for the job and continue to the next screen.
  04:28	The "Configure" step allows you to review what the input and output of your job run will be.
  04:36	And select the environment used to run the job.
  04:41	Scheduling a job is optional, but you can set a date and repeat the job, if you'd like.
  04:51	And you can choose to receive notifications for this job.
  04:56	Everything looks good, so create and run the job.
  05:00	This could take several minutes, because remember that the data flow will be run on the full data set.
  05:06	In the mean time, you can view the status.
  05:12	When the run is compete, you can go back to the "Assets" tab in the project.
  05:20	And open the Data Refinery flow to further refine the data.
  05:28	For example, you could sort the "AverageDelay" column in descending order.
  05:36	Now, edit the flow settings.
  05:39	On the "General" panel, you can change the Data Refinery flow name.
  05:46	On the "Source data sets" panel, you can edit the sample or format for the source data set or replace the data source.
  05:56	And on the "Target data set" panel, you can specify an alternate location, such as an external data source.
  06:06	You can also edit the properties for the target, such as the write mode, the file format, and change the data set asset name.
  06:21	Now, run the data flow again; but this time, save and view the jobs.
  06:28	Select the job that you want to view from the list and run the job.
  06:41	When the run completes, go back to the project.
  06:46	And on the "Assets" tab, you'll see all three files:
  06:51	The original.
  06:54	The first refined data set, showing the "AverageDelay" unsorted.
  07:02	And the second data set, showing the "AverageDelay" column sorted in descending order.
  07:11	And back on the "Assets" tab, there's the Data Refinery flow.
  07:19	Find more videos in the Cloud Pak for Data as a Service documentation.

1. Access Data Refinery from within a project. Click New asset > Prepare and visualize data. Then select the data that you want to work with. Alternatively, from the Assets tab of a project, open a file ( supported formats) to preview it, and then click Prepare data.

Note that different asset types can have duplicate names. However, you can't add an asset type with the same name multiple times.

2. Use steps to apply operations that cleanse, shape, and enrich your data. Browse operation categories or search for a specific operation, then let the UI guide you. You can enter R code in the command line and let autocomplete assist you in getting the correct syntax. As you apply operations to a data set, Data Refinery keeps track of them and builds a Data Refinery flow. For each operation that you apply, Data Refinery adds a step.

Data tab

If your data contains non-string data types, the Convert column type GUI operation is automatically applied as the first step in the Data Refinery flow when you open a file in Data Refinery. Data types are automatically converted to inferred data types, such as Integer, Date, or Boolean. You can undo or edit this step.

3. Click the Profile tab to validate your data throughout the data refinement process.

Profile tab

4. Click the Visualizations tab to visualize the data in charts. Uncover patterns, trends, and correlations within your data.

Visualizations tab

5. Refine the sample data set to suit your needs.

6. Click Save and create a job or Save and view jobs in the toolbar to run the Data Refinery flow on the entire data set. Select the runtime and add a one-time or repeating schedule. For information about jobs, see Creating jobs in Data Refinery.

For the actions that you can do as you refine your data, see Managing Data Refinery flows.

Next step

Analyze your data and build models

Learn more

• Manage Data Refinery flows

• Quick start: Refine data

Parent topic: Preparing data