Using Stratifyd in Tableau

You must first have a Stratifyd dashboard to use in Tableau. For details on how to create one, see Creating a dashboard.

You must also have a web data connector set up to connect to Stratifyd. Ask your Stratifyd contact about setting it up for you.

To connect to data

1. Open Tableau, and in the Connect pane to the left, under To a Server, click More... 

Alternatively, if you already have a workbook open, in the Data pane, click Connect to Data.

2. In the expanded menu, select Web Data Connector. 

3. In the Web Data Connector dialog that appears, enter the URL of your Stratifyd platform Tableau web data connector page (twdc.html), e.g. https://YourCompany.stratifyd.com/twdc.html, and press Enter. 

triangle: Your server administrator must add the URL for the connector to the safe list on the server in order to refresh the data. 

More information from Tableau help:

• If you open a workbook on Tableau Server that was created using a web data connector, but the connector has not been added to the safe list on Tableau Server, and you want to be able to refresh the extract on Tableau Server, follow the process for testing, vetting, and adding the connector to the safe list. If the connector requires credentials to sign in, you need to ensure that the credentials are embedded with the data source. You can then refresh your data on Tableau Server.

• When you publish to Tableau Online, as a security measure, Tableau Online can't connect to or refresh an extract created by a web data connector. To refresh some web data connector extracts, you can use Tableau Bridge. For more information, see When to use Tableau Bridge to keep data fresh in the Tableau Online Help.

• When you publish to Tableau Public, because you can't add a web data connector to Tableau Public, you can't refresh web data connector extracts directly on Tableau Public].

4. In the Stratifyd Login screen that appears, enter your Stratifyd user name and password and click Log in.

Logging in to Stratifyd via Tableau automatically logs out any Stratifyd browser session that may be open using the same credentials. Likewise, logging into a Stratifyd browser session with the same credentials automatically logs you out of the connection in Tableau.

5. Once logged in, click the Select a Dashboard to Export button. 

6. In the Pick a dashboard list that appears, click the Stratifyd dashboard to use as a Tableau table. 

7. In the list of Export Analyses that appears, you can see any data streams that are included in the dashboard, and under each data stream, you can see any data models that have been created from the data and trained within Stratifyd. 

Click to select a supported data model (Unsupervised NLU or Taxonomy) and in the Export unique ID box that appears, click to the right of the text and choose the field that contains your unique identifier in the Tableau dashboard.

 An X to the left of a data model indicates that it is not supported in Tableau, while an O indicates that it is supported in Tableau.

8. Skip this step if you only have one data model in the Stratifyd dashboard, as it automatically appears on the canvas.

On the data source page that appears, Stratifyd is listed under Connections. Under Table, if you have multiple tables (Stratifyd data models), drag the one that you want to use and drop it onto the canvas. 

9. To join metadata to the model output, in the Connections panel on the left, click Add, and select your metadata source. 

10. To join the tables, under the Stratifyd Data Source column, select Unique ID. 

11. Under your metadata source column, choose the corresponding ID field to complete the join. 

The joined circles between the data sources indicate a successful join. 

12. Column headings from the table appear below the canvas. Click Update Now to preview your joined data. 

Depending on the amount of data, this may take some time. You can skip this step if you already are familiar with the data source.

For example, with 7,000 records, it takes approximately two minutes. Your results will vary based on internet connection speeds and processor speeds.

The table populates with data that you can use to create visualizations. 

13. Now you can use the data to create visualizations in Tableau as you normally would, but with the advantage of machine learning insights.

Supported data models

Unsupervised NLU model

The Unsupervised NLU (natural language understanding) model automatically analyzes large textual datasets to discover context-based semantic topic groups and themes in your data. Built on top of our proprietary Bayesian Neural Network and Generative Model, it categorizes sentiment without human bias, and the machine learning functionality improves with run time.

Each document or record may have multiple buzzword pairs, so you may see a number of rows in the table with the same Verbatim or Review text.

Taxonomy

What is a taxonomy?  

Stratifyd uses faceted hierarchical taxonomies with controlled vocabularies. Labels are broad or narrow branches of the hierarchical categorization tree. The controlled vocabulary for each label is a restricted list of words and Boolean logic used to index documents.

Controlled vocabularies in Stratifyd do not generate synonym rings. You add your own custom synonymous terms to the vocabularies for each label. Documents are funneled into labels based on whether they "did match" or "did not match" one of the terms in the label's controlled vocabulary. Narrower labels work with their broader parent labels' vocabularies as nested levels of matching logic, enabling a more fine-grained classification model. 

The taxonomy table has columns like the following when you connect to them in Tableau. 

 Each document or record may have multiple taxonomy labels, so you may see a number of rows in the table with the same text.