Company Blog

Kicking Off 2019: The Year of Model Operations

kicking off 2019Welcome to 2019 – a year we here at ODG are calling “The Year of Model Operations”.  Throughout the year, we will be bringing a wide array “how to’s”,  customer and partner case studies and use cases – examples of what it means to execute machine learning strategies into production.  It’s an exciting time, as we’ve seen the shift of the market from “what is model operations” to “how do I do model operations”.

 

I wanted to start with a look back at three important articles published by Gartner to set the stage for what’s to come.  The first article was published in Q4, by Kasey Panetta, titled “Gartner Top 10 Strategic Technology Trends for 2019”, on gartner.com.  The other two research notes were published by Erick Brethenoux and others also in Q4, titled “Predicts 2019: Analytics and BI Strategy” and “How to Operationalize Machine Learning and Data Science Projects”.  It’s a combination of these three that are particularly interesting.

 

In Kasey’s article, she outlines some very powerful secular trends that will see significant implementations in 2019.  Of the 10 included, three are very specifically related to the adoption and implementation of machine learning and AI:  Augmented Analytics, AI-Driven Development and the Empowered Edge.  Each of these trends, at a very fundamental level, will rely on a host of ML and AI models to provide value.  Importantly for any of these trends to be highly adopted, it implies those ML and AI models will be deployed and supported in production, a topic we call model operations.  In contrast, let’s look briefly at Erick’s article from earlier last year.  In his research, Erick notes “More than half of Data Science Projects are not Fully Deployed”.   The fact that many ML projects are not deployed implies there is a large gap looming to achieve some of the key trends identified by Kasey.  In addition, Erick and team note in their 2019 Predictions that organizations “Do not deploy analytic tools without first ensuring that processes are established to drive operationalization of the results, and that the business culture is receptive to analytic insight. “

 

Based on our customers, and the work we have in the Fortune 500, it’s clear this model deployment gap is real.  Importantly, the gap exists in complete alignment with the Gartner research:  there is both a process gap, and a technology gap.  We’ve been engaged by banks, insurance companies, retailers and many others to close this gap, and help them gain real competitive advantage and leadership positions in leveraging ML to impact their business. 

 

As a provider to the space, what’s exciting is that the market, regardless of customer or use case, exhibits common patterns that cause these challenges.  And if we can find common patterns, we can solve for them.  A few of these patterns have emerged, and 2019 is indeed the time to solve for them:

  1. Cloud infrastructure is here to stay, and teams need to be able to deploy ANY machine learning model into any infrastructure chosen, and take advantage of the cost and flexibility the cloud provides.
  2. Data science, quant, and analytics teams must be free to choose any combination of model creation tools, languages and packages they need to create the best model for their use cases.
  3. Enabling critical mathematical assets to be operated safely, at scale, by traditional DevOps teams is required to put models into production.
  4. Processes to support model operations, including a Model Development Life Cycle (MDLC), are required to support the larger enterprise vision of transformation via data and ML.

 

2019 is the year of the model operations.  The good news is that we’ve seen our customers solve the problems, generating tremendous value from their ML investments.  Combining FastScore and the MDLC approach solves for fundamental model operations challenges, enabling several of the key trends that Gartner, and our customers, are aiming for.  Over this year, the team here at ODG will be working to share in depth “how you do it” examples, all based on our experience with our customers. Please stay tuned for both new content and customer stories to come.

All ModelOp Blog Posts 

Machine Learning Model Interpretation

To either a model-driven company or a company catching up with the rapid adoption of AI in the industry, machine learning model interpretation has become a key factor that helps to make decisions towards promoting models into business. This is not an easy task --...

Matching for Non Random Studies

Experimental designs such as A/B testing are a cornerstone of statistical practice. By randomly assigning treatments to subjects, we can test the effect of a test versus a control (as in a clinical trial for a proposed new drug) or can determine which of several web...

Distances And Data Science

We're all aware of what 'distance' means in real-life scenarios, and how our notion of what 'distance' means can change with context. If we're talking about the distance from the ODG office to one of our favorite lunch spots, we probably mean the distance we walk when...

Communicating between Go and Python or R

Data science and engineering teams at Open Data Group are polyglot by design: we like to choose the best tool for the task at hand. Most of the time, this means our services and components communicate through things like client libraries and RESTful APIs. But...

An Introduction to Hierarchical Models

In a previous post we gave an introduction to Stan and PyStan using a basic Bayesian logistic regression model. There isn't generally a compelling reason to use sophisticated Bayesian techniques to build a logistic regression model. This could be easily replicated...