5 Takeaways from the 2022 Gartner® Data & Analytics Summit, Orlando, Florida
How do you drive collaboration across teams and achieve business value with data science projects? With AI projects in pockets across the business, data scientists and business leaders must align to inject artificial intelligence into an organization. At the 2022 Gartner Data and Analytics Summit, data leaders learned the latest insights and trends. Here are five key takeaways from one of the biggest data conferences of the year.
Data Analysis Must Include Business Value
To drive business value and successfully apply AI, it’s critical that members of data and analytics teams clearly articulate the underlying business value. Not only is this a requirement, it needs to happen at project kickoff, rather than waiting until the end. While this may not be groundbreaking in concept, storytelling skills are not always innate for some individuals.
That’s why DataRobot University offers courses not only on machine learning and data science but also on problem solving, use case framing, and driving business outcomes. Because it’s not just about the data itself, it’s about how you convey the value and solve use cases. DataRobot Solution Accelerators help further speed up the process by providing a quick starting point.
Collaboration Matters Across the AI Lifecycle
Whether it’s decision thinking or driving innovation, working in silos is not a good option for today’s organizations. Data science teams cannot create a model and “throw it over the fence” to another team. Everyone needs to work together to achieve value, from business intelligence experts, data scientists, and process modelers to machine learning engineers, software engineers, business analysts, and end users. Repeatedly, the phrase “AI is a team sport” needs to be reinforced across the business, as stated by Gartner analyst Arjun Chandrasekaran.
DataRobot has unified the experience for all users within a single platform. With an intuitive interface and out-of-the-box components, you can reach your goals and be efficient without deep data science expertise or coding skills. At the same time, advanced data scientists interested in experimenting or bringing their own models and leveraging automation can easily do this, too. And lastly, engineers managing IT or production environments find it simple to connect the DataRobot AI platform to other tools.
Transparency Is Key In MLOps
While collaboration is critical to success, it also introduces challenges with visibility. This becomes increasingly important as more teams across an organization develop models. As mentioned by Gartner analyst Sumit Agarwal in his session, Developing Your MLOps Playbook to Accelerate Machine Learning Deployment, “one person cannot do everything.”
Model observability is more and more critical, especially in fast-changing environments. Having complete visibility gives you control over your production AI. With powerful built-in insights, you can quickly evaluate, compare, and decide about model replacement. You can also go beyond regular accuracy and data drift metrics. With custom metrics, you can access your training and prediction data and implement any metrics that are relevant for your business case.
Perfection Is the Enemy of Progress
While accuracy is important, we are too often stuck in the mindset of achieving perfection at the expense of forward momentum. Often, good enough is the best route. An additional month of missed opportunity means unrealized value for the business. Knowing what is good enough is a critical skill for individuals leading AI projects. The term Gartner uses for this is “satisficing” – focusing on continuous improvement.
The end-to-end experience of the DataRobot AI platform allows you to experiment fast and get your first model into production. Then, as your model gets deployed, you can set up challenger models that will work in a shadow mode with different parameters. With the Challengers framework, you can always have options to choose from to ensure that you have top performing models in production. In addition to model challengers, automated retraining reduces the amount of manual work to retrain a model.
Interoperability Extends the Impact of AI
The goal with data science and machine learning is to inject AI into the DNA of an organization. To do this, an AI platform needs to be flexible and extend into other systems, allowing AI to be pervasive and removing barriers to adoption.
Built as a multi-cloud platform, DataRobot AI Platform enables organizations to run on a combination of public clouds, data centers, or at the edge, with governance to protect and secure your business. It is modular and extensible, building on existing investments in applications, infrastructure, and IT operations systems. DataRobot AI Platform is powered by a global ecosystem of strategic, technology, solution, consulting, and integrator partners, including Amazon Web Services, AtScale, BCG, Deloitte, Factset, Google Cloud, HCL, Hexaware, Intel, Microsoft Azure, Palantir, Snowflake, and ThoughtSpot.
Gartner, Technical Insights: Develop Your MLOps Playbook to Accelerate Machine Learning Deployment, Sumit Agarwal
GARTNER is the registered trademark of Gartner Inc., and/or its affiliates in the U.S. and/or internationally and has been used herein with permission. All rights reserved.