Winners and Losers from Gartner’s Data Science and ML Platform Report
Gartner published its latest Magic Quadrant for data science and machine learning platforms last week. Sixteen vendors made cut for Gartner’s report this year, the same number as last year. However, there were some important changes, including some vendors who made big jumps and some who lost ground.
The biggest difference arguably was the addition of “machine learning” to the name of Gartner’s report. “Although data science and machine learning are slightly different,” the Gartner analysts write, “they are usually considered together and often thought to be synonymous.”
With that said, here are the winners and losers from Gartner’s Magic Quadrant for Data Science and Machine-Learning Platforms for 2018, and those who maintained their ground:
H2O progressed from the Visionaries Quadrant last year into the Leaders Quadrant. In fact, H2O scored the highest in terms of its completeness of vision out of all the vendors. Gartner lauded H2O’s support for cutting-edge tech, like deep learning and hybrid-cloud support, and for how it’s become a “quasi-industry standard.” High customer satisfaction also boosts H2O’s score. Cautions include a primarily code-centric approach (not for the point-and-click crowd), a lack of compelling data prep and visualization features, and a “give it all away” business model that’s entirely dependent on revenue from tech support.
Alteryx moved up from the Challenges Quadrant to the Leader’s Quadrant, one of the biggest jumps in this year’s report. In its review, Gartner liked how Alteryx is building more automation and rules-based recommendations into its platform to help customers operationalize data science insights, but wondered where automated model building capabilities were. The analyst group also cited the company’s revenues and customer satisfaction (it didn’t mention the large data breach that impacted Alteryx last year several after it went public). Drawbacks of the firm include a perception that Alteryx is just for data preparation (it’s not) and enterprise readiness.
Anaconda (formerly Continuum Analytics) debuted in the Niche Players quadrant. Gartner seemed impressed by Anaconda’s capability to federate Python-based ML model building for large numbers of developers. It liked the open source aspect of Anaconda’s work, and how Anaconda is supporting and even indemnifying open source Python-based data science libraries. But it raised a couple of flags, including a minimum of automation in the product and documentation. Beginning Anaconda users, Gartner says, “will have difficulty finding their way through the Python ‘jungle.'”
Databricks debuts in the Visionaries Quadrant on the power of its cloud-based offering based on Apache Spark. Since Databricks spearheads the development of Spark, its technical bona fides are unquestioned, particularly for working with large data sets. Gartner also admired Databricks flexibility, as evidenced by its work in streaming, deep learning, and Internet of Things (IoT) areas, not to mention supporting multiple languages (R, Scala, Python) with its notebook-based development environment. Caveats to Databricks’ approach include market awareness, visibility of cost associated with the underlying third-party cloud platform, and debugging.
Domino Data Lab finds itself once again a Visionary in Gartner’s quadrant, where it has improved its standing somewhat and is now clustered right up against two tech behemoths, IBM and Microsoft. Gartner was impressed by Domino’s flexibility, extensibility, openness, and lauded the company’s tech support and how it incorporates user’s ideas into the product. Caveats include a “high technical bar,” some difficulty getting started with ML projects, a lack of “precanned” solutions, and enterprise-grade functions.
KNIME retains its spot in the Leader’s Quadrant, a spot it has occupied for quite some time, in both this quadrant and Gartner’s advanced analytics reports. Gartner is impressed with how the Swiss firm delivers a unified and consistent data science experience, and how it integrates with R, Python, Spark, H2O.ai, Weka, DL4J and Keras. Automated model development and deployment earned it high marks, while low cost earned it more admiration. Drawbacks of the KNIME approach focus on a lack of marketing and sales innovation, as well as lack of commercial options (KNIME is all open source). Performance and scalabiity, particularly with large data sets, was also cited.
IBM lost a ton of ground in Gartner’s Magic Quadrant. It previously occupied the sweet spot in the quadrant — farthest up and to the right — but Gartner dinged Big Blue substantially across both metrics, completeness of vision and ability to execute. Gartner evaluated two SPSS products for its recent review; its Data Science Experience didn’t meet the qualifications. Strengths include IBM’s market share (9.5% of the entire data science market, Gartner says), generally strong analytic capabilities for experts and novices alike, strong data preparation and model management, and a legacy user base. Cautions include a lack of a unified offering, the transition to DSX, and general confusion about Watson. High cost and a poor customer experience also hurt IBM.
Teradata is still in the Niche Player’s category, but it fell substantially and features the lowest ability to execute out of all 16 vendors featured. Gartner liked the performance and scalability offerd by the San Diego, California, analytic firm’s products. It got high marks for supporting existing data structures, for having pre-canned solutions, and for having an army of consultants. But it was dinged for model development in the Unified Data Architecture,, pricing and platform coherence.
Angoss suffered a deterioration in its “ability to execute” according to Gartner and was demoted from the Challenger’s quadrant into the Niche Player’s quadrant. The firm, which was acquired by DataWatch in January, develops an easy to use and well-rounded desktop data science product, Gartner says. Its embrace of open source tools like Spark, TensorFlow, and the libraries from H2O bolster the credentials for the KnowledgeSEEKER and KnowledgeSTUDIO products. But market traction, a desktop-only perception, and limited industry use (it’s mostly used in banking) make up the cautions.
RapidMiner repeated as a member of the vaunted Leader’s Quadrant, thanks to its “well-rounded and easy-to-use” platform. Gartner says RapidMinder’s software, which is mostly open source, offers “deep and broad” modeling capabilities, and it was particularly impressed with how it allows retraining of models on new data. Caveats include an ongoing tendency to add closed-source components to its commercial product. Data visualization and documentation are two areas it can work on, Gartner says.
SAS remains in the Leaders Quadrant, but lost some ground in the completeness of vision and ability to execute categories. The analyst group liked the North Carolina firm’s Visual Analytics product and its open Viya architecture, and lauded its overall “operational excellence.” But it faulted SAS for its marketing and sales execution, its complex and multi-pronged product strategy, and its product and sales strategy.
Microsoft lost some ground this year. It was on the cusp of moving up to the Leader’s quadrant last year, but now finds itself in the middle of the pack in the Visionaries quadrant. The company’s ability to offer products, notably Azure Machine Learning Studio, on the cloud is obviously a big plus in Microsoft’s book, as are its marketing execution, mindshare, and support. It also sports a “visionary roadmap,” according to Gartner, which perhaps explains its listing in this quadrant. Drawbacks include a cloud-only approach, a lack of data preparation capabilities, and market responsiveness.
SAP moved slightly in the Niche Players category for its SAP Predictive Analytics platform, which previously was marketed under the Business Objects brandf. Gartner gave SAP high marks for collaboration, its unified Leonardo machine learning capability, and “some strong product capabilities.” But it dinged the German software giant for its fragmented and ambiguous toolchain, and its slow adoption of cognitive technologies, and its approach to cloud.
TIBCO Software debuted in the Challenger’s Quadrant, largely on the back of two 2017 acquisitions: Statistica and Alpine Data. Gartner lauded the firm for its model management and operationalization capabilities, as well as its support for IoT use cases and its customer support. Lower marks were recorded for the number of times Statistica has changed hands (four times in four years), as well as for cloud capabilities, end-to-end integration, and performance and scalability.
MathWorks, the developer of MATLAB, lost a bit ground. It was solidly in the Challenger’s quadrant last year, and now finds itself on the edge of falling into the Niche Player’s quadrant. The analyst group cites a large installed base that’s quite loyal as a big plus for MathWorks, particularly in areas like engineering, manufacturing, and quantitative finance. Negatives include a lack of focus for other use cases, like marketing, sales, or customer service; a lack of focus on incorporating open source technology; and poor customer feedback on the company’s willingness to incorporate requests into the product.
Dataiku finds itself in the Visionary Quadrant, the same place it was last year. However, it lost ground along both axes. Gartner remains impressed by the overall ease-of-use of working with the Dataiku DSS platform, and how quickly users can get started with prototyping machine learning projects. It also likes the “pluggable” architecture that lets users incorporate models developed in other environments (like Spark MLLib and the ever-popular H2O offering). Cautions include the process of moving from development into production, a lack of automation, and pricing.