A community thrives on the basis of its great members - so let's take a look at some of them shall we? From time to time we will be highlighting a prominent member of the Community, and sharing their story and DSS accomplishments!
Meet Ben - known here as @ben_p. We sat down with him and had a chat about his story with Dataiku DSS.
1) How did you find Dataiku and get started with DSS?
I bumped into Dataiku at a couple of conferences in 2018/19. At this time We had a couple of models in deployment, as .py files running on virtual machines, and this was working well. When my team started to deploy more models we quickly realized that our current approach wasn't very scalable, and that actually maintaining models without much automation was creating a lot of manual work.
We identified a need for an ML platform, to help us with deployment and all the other steps of the model lifecycle. We went to market and assessed a lot of different solutions, eventually selecting DSS - we were really impressed with the suite of tools available and particularly the way DSS makes it really easy for a team to work collaboratively on projects and share code.
We brought DSS into the business in January 2020 and within 8 weeks we had migrated all of our daily ETL processes into the platform, benefiting from automated checks and triggers, making our process much more robust!
2) What's your favorite DSS feature?
I have to pick one? I love the data cleaning functions - being able to parse a string date then break it down into component date-parts in a few clicks has already saved us so much time.
Also (because I am going to choose two!) I love the way DSS makes it so easy for a team to work together. These collaboration tools have really sped up our workflows and allowed team members to work more independently.
3) Tell us about your projects!
We have one project which handles all of our daily ETL processes, running a lot of SQL jobs and Python scripts. Being able to represent this visually, in a flow, and set dependencies and order has really helped up streamline a huge number of jobs into a reliable process.
As an online retailer we do a lot of modeling on our customers, most recently we have delivered a Customer Lifetime Value model with DSS, which looks at many customer features and predicts the revenue we can expect to receive from each in the next year.
Also in the customer space, we have worked on propensity modeling; one example is customer return propensity, where we look at all our website visits from the previous day and all the interactions our visits made with the site. We run these through a model and output a score showing the likelihood that each visitor will return to the site in the next few days.
Our models are typically deployed (on a DSS automation node) to make batch predictions, but we are really excited to check out the new application features in DSS. We want to share access to our models with non-technical colleagues around the business and have creating predictions in real time!
Do you have a Dataiku story? Share how you came to use Dataiku DSS or an interesting goal you accomplished with it!