How to deploy your Analytics Project: Use Cases

Details
Hi Deployment enthusiasts,
Sponsored by GoDataDriven, this month we’ve invited two guest speakers to delve deep into Deployment and Production on July 12th!
First Talk: "Predicting and deploying Dutch House prices"
When you want your model to be meaningful and have an impact on 'day-to-day' operations of a business, it is necessary to deploy your predictive models into production so that others can use it. One way to operationalize your model is to expose your model through a REST API.
To create such a REST API for a predictive model and to set it up so that it easily scales, is secure and is manageable if you have many models, can be very time-consuming. A time that some of us prefer to spend more on understanding the data better and talking to the business :-). Luckily the new version of Dataiku can speed up this process and help you out.
In this talk, Longhow will explain how he modeled and deployed a house price model for the Dutch market.
Longhow Bio:
Longhow Lam is a freelance data scientist. Currently he is working at the SVB, the organization that implements national insurance schemes in the Netherlands. He is involved in text mining documents at the SVB to improve customer contact and to use data science to combat fraud.
For two years he is an enthusiastic Dataiku user, and in his spare time he does some 'frivolous' hobby data analysis which you can read on his LinkedIn timeline. https://www.linkedin.com/in/longhowlam/detail/recent-activity/posts/
Second Talk:
"Testing and monitoring for production-ready models"
Deploying a machine learning model to production is a crucial step in a data science project, but it's definitely not the final one! For an ML system to reach top levels of efficiency, there are two main topics in a model lifecycle management that need to be thoroughly understood:
- which metrics you need to monitor to make sure the model is still relevant in the production context
- which tests you want to perform at deployment and serving time to assess the robustness of your model
The goal of this talk is to dive deeper into those topics and to bring some
elements of answers to the question of monitoring deployed models, with some implementation examples on Dataiku DSS.
Bio:
Harizo Rajaona is a data scientist in the EMEA team of Dataiku. He has been collaborating with several clients to help them push their data science projects to production, and has been recently focusing on the
operationalization of ML models.
A message from GoDataDriven:
https://godatadriven.com/data-engineering-training-program
Are you an experienced IT professional with an interest in data and machine learning in the cloud? Would you like to learn how to take scalable machine learning products into production for leading Dutch organizations?
GoDataDriven, the leading data engineering service provider, has a unique opportunity for five select IT professionals to develop data engineering proficiency. Our tailor-made training program results in a guaranteed position as a data engineer, but there are only five spots available.
Interested? Send us your CV and tell us why you want to be part of our team. We’d love to have you onboard! career@godatadriven.com

How to deploy your Analytics Project: Use Cases