Databricks – pricing-frontier.co.uk

What is Databricks?

Databricks is an all-in-one data engineering and data science platform, it is cloud based and accessible through a web browser.

We have partnered with Databricks because we believe it is the easiest way for Pricing teams to adopt modern analytical practices.

How does it work?

The platform uses code (Python, R, SQL or Scala) for data engineering, data science and data analytics processes.

The code is written in notebooks and run on clusters of machines, this allows tasks to be split into partitions and run simulatenously, allowing large datasets to be processed quickly.

Databricks offers a storage solutions, which are designed to make working with data easier, efficient and robust.

How can it help Pricing teams?

Databricks has everything needed to adopt modern analytical practices:

Notebooks can be referenced within others, allowing tools and scripts to be built in a single place, and reused across the codebase; reducing complexity and duplication of processes.
Notebooks can be scheduled to run automatically, allowing up-to-date data for analysis, modelling and reporting to be readily available at all times.
There is version history of notebooks, and version control. This allows for a production codebase, and analysts can create branches to work on separate branches simultaneously and merge these to the main when ready.
Interacts well with open-source technology allowing Pricing team flexibility as they are not limited to closed-source software.

The machine learning runtime has the most common libraries installed (and any additional ones can be easily added). Version control of models is built in and models can be easily be deployed to an endpoint that allows other tools to query them.

Can Pricing teams do this without Databricks?

These practices could be adopted with open-source tools. The benefit of Databricks is that all the infrastructure is managed, and so does not require installation and setup of a multitude of tools and services and keeping these consistent across users. Creating a cluster is as simple as choosing how big it needs to be, and onboarding analysts is a case of adding permissions to their logins.

What does Pricing Frontier offer?

We have been using Databricks for several years and have experience building a range of different data science solutions using the platform.

Combined with Insurance Pricing expertise, we have been able to build Pricing architectures that significantly improve the efficiencies of Pricing teams; these have been built to be easy to understand and maintainable. Examples of efficency gains:

Developing data pipelines that produce data for all pricing processes including analysis, modelling and reporting.
Automating and streamlining very time consuming tasks, significantly reducing time required.
Machine learning training, deployment, monitoring and refreshing.
Automated reporting and A/B tests.

We understand the modern data landscape and have helped Pricing teams upskill in this area and utilise the tools within Databricks.

Further information

You can find more information on the Databricks website or get in contact for a free consultation on modernising your Pricing team.

Pricing Frontier