Converting your Dataiku DSS Project into a Reusable Application - Watch on Demand

Dataiker Alumni
1 min read 4 4 2,760

Applications are an exciting new Dataiku feature which empower a wide variety of people within an organization to easily leverage AI and self-service analytics. In this session, @rmoore (Vice President of AI Solutions at Excelion Partners) shows you how easy it is to convert a typical Dataiku project into a reusable Application that can be utilized throughout the enterprise.  


Ryan was also kind enough to show a link to the project itself which can be accessed via GitHub.

Before the event check out this resource: Difference Between Webapps and Dataiku Applications

Be sure to join the Chicago User Group to be informed of upcoming events and chat with fellow DSS users based in the Chicagoland area!

As you may know, Dataiku User Groups are led by volunteer users who contribute their time and communication skills to enable fellow users to learn from each other. If you’d like to run this group, please fill out this quick form and we’ll get back to you. Thanks for your interest!


Thanks for the great presentation Ryan!

How often do you find yourself having to update the design of your applications due to requests from the users (e.g., perhaps requests for additional charts, or more input parameters, or even additional variables)?



Hi @taraku - great question! We generally view Data Science projects and applications as "agile" in nature, therefore in a pretty constant state of evolution. 

We do try to push our users for consistent feedback so we can update the applications as you're suggesting and tailor them to best suit the business needs. 

Let me know if I can provide further details!


I've also posted the Market Basket Application project I mentioned:

Please let me know if anyone has any issue getting it running.

@tgb417 @CoreyS 




Thank you so much for sharing this Dataiku Application.  I'm very interested in Applications as a way of creating reproducable data science.

I'm sharing the following for others trying to reproduce this work in their dss environment.  Note this only going to work for Versions of DSS >=8.0.0 (I'm not clear if this has to be 8.0.2 or later).

When I tried to import the project from a downloaded copy of the GitHub repo. The .zip file seems very small for a project file at 2kb.

When trying to install this as a project file, I'm getting the following error message:

Market Basket Error.jpg


Clearly, that did not work, as I've investigated further.  It's become clear that this is not a project file, this is a code environment import file.  Ooops.

However, when I try to import as a code environment, I'm getting the error.



So, I've gone and created a code environment as instructed on the GitHub repo:

If this does not import correctly, create a new Code Environment
called market_basket_app using Python 3.6 and add the following
PIP packages:
- efficient-apriori
- dataclasses

Thanks that seems to work.

The file is 7Mb.  That's more like it when it comes to a project file.

While importing the project file I got the following warnings.

Import Error.jpg


These are all only warnings.  We will see how this goes.  I suspect based on what is listed here that things will be OK.

I was able to build the data in the underlying project with out problems.

However, when attempting to look at the dashboards I got lots of errors.  Hmmm...

Dashboard Errors.jpg

To get this to work I had to go to the Recommendations_per_Customer dataset in the follow and refresh the statistics.  Now that I've done that I'm seeing data.  Good.

However, I'm having problems getting the dashboard to refresh when I run the application with different parameters.  Like different countries, and support and confidence levels  The data is always coming out the same.