David Menninger's Analyst Perspectives

Machine learning is valuable for organizations, but it can be hard to deploy. Our Machine Learning Dynamic Insights research identifies that not having enough skilled resources and difficulty building and maintaining ML systems are pressing challenges organizations face in applying ML. Traditional ML model development is resource-intensive, requiring significant domain knowledge and time to produce and compare dozens of models. And as the number of ML models grow, their management becomes difficult. By bringing automation to ML, organizations can reduce the time it takes to create production-ready ML models. AutoML can also enable organizations to make data science initiatives more accessible across the organization.

DataRobot is an automated ML platform that can enable users of various skill levels ̶  from business personnel to analysts to data scientists ̶  to build and deploy ML models. It offers an intuitive visual interface that works for both line-of-business personnel and experienced data scientists. The DataRobot platform uses parallel processing to train and evaluate models in R, Python, Spark MLlib, H2O and other open-source libraries. It can search through millions of possible combinations of algorithms, pre-processing steps, features, transformations and tuning parameters automatically. DataRobot also provides a library of algorithms and pre-built prototypes for feature extraction and data preparation.

DataRobot recently released the 7.0 version of its AI/ML platform, introducing upgrades to much of their portfolio, including image augmentation to Visual AI, and customizable compliance reports in AutoML. One of the major enhancements includes MLOps remote model challengers, which allow organizations to analyze how challenger models perform against the current top production model to evaluate how they would have performed during the most volatile moments in history. This can help inform decisions about whether to keep or replace the current champion model and make more accurate predictions. Another enhancement is prediction preparation that allows organizations to prepare their data for model training. Business personnel can use visual data prep to score new data from models already deployed.

As highlighted in the chart above, data preparation is also a challenge of AI/ML efforts. DataRobot Data Preparation is a self-service data preparation platform that enables data engineers, data scientists and business personnel to apply complicated data preprocessing techniques through an easy-to-use graphical user interface. Paxata can also create an automatic project flow to operationalize data flows, and can automatically process the entire sequence of data preparation steps across multiple Paxata projects and datasets.

Organizations are realizing the benefits of AI/ML in extracting business value from data. Through 2022, artificial intelligence and machine learning solutions will remain largely independent of business intelligence solutions, requiring three-quarters of organizations to maintain multiple, separate skill sets. Organizations will need a platform such as DataRobot’s to build and deploy AI/ML models. With DataRobot, people with different skills ranging from data scientists to business analysts to IT can collaborate in extracting business value from data with AI/ML.

DataRobot MLOps provides tools to deploy, monitor, manage and govern models in production. MLOps also offers built-in, write-back integrations to systems such as Snowflake and Tableau. DataRobot Visual AI provides the ability to include images in ML pipelines, allowing both novices and experts to build and deploy ML models using image-based data. DataRobot also offers the functionality to convert any ML model (DataRobot-generated models or models written in R or Python) into an AI application. This enables business personnel to interact with the predictive insight of the underlying model and experiment with different scenarios, predict results and make more informed business decisions by interacting with the application.

Organizations are discovering the business benefits of machine learning and predictive analytics, but they are also discovering that data science talent is hard to find, costly to acquire and challenging to retain. Additionally, most ML deployment processes today are manual, complex and span data science, business and IT organizations, impeding the rapid detection and repair of model performance problems. DataRobot is enabling data scientists and analysts to produce more value by making ML simpler to use and deploy. In addition, DataRobot should continue to invest in governing AI/ML models, and add further integration capabilities with analytics and BI tools. DataRobot can also add more write-back capabilities to support a wider variety of databases and business applications.

Organizations looking to empower an ecosystem with ML should consider the capabilities of DataRobot when evaluating vendors. The DataRobot platform offers a straightforward interface for skilled data scientists as well as non-data scientists to run and deploy models in an automated way.

Regards,

David Menninger