Machine Learning, The New Frontier In Data Science
The ability to draw insight from massive streams of data is becoming a competitive differentiator for enterprises. Gartner evaluated software vendors offering products that allow development and deployment of the data science workloads that deliver that insight.
The companies in this Magic Quadrant bring to market "a cohesive software application that offers a mixture of basic building blocks essential both for creating many kinds of data science solutions and incorporating such solutions into business processes, surrounding infrastructure and products."
Machine learning is a subset of the technology, but one that warrants specific attention. Machine learning's momentum and its significant contribution to the broader discipline of data science is why the term, for the first time this year, has been added to the title of this Magic Quadrant.
The 16 platforms evaluated by Gartner support data scientists doing work across the analytics pipeline. Those projects have many components, including data access and ingestion, data preparation, interactive exploration and visualization, feature engineering, advanced modeling, testing, training, deployment and performance engineering.
The Gartner Magic Quadrant ranks vendors on two criteria: Ability to Execute and Completeness of Vision.
Execution is represented on the Y-axis of Gartner's chart; vision on the X-axis. That lands competitors into one of four quadrants: Niche Players (low in both criteria), Visionaries (complete vision but lacking execution), Challengers (good execution but lacking vision), and Leaders (excelling in both vision and execution).
Among the 16 software developers evaluated this year, only five made the Leaders quadrant. Four were relegated to Niche Players, two just edged out of that quadrant and into the Challengers one, and five were ranked as Visionaries. Following are each vendor's rank as well as what Gartner had to say about its offering.
The Mountain View, Calif., software company is Gartner's clear leader in Completeness of Vision.
H2O.ai is an open-source machine-learning platform that scores well in its deep-learning and automation capabilities, hybrid cloud support and open-source integration.
"It continues to progress through significant commercial expansion, and has strengthened its position as a thought leader and an innovator," Gartner noted.
The company's technology abstracts away many of the details of deep-learning frameworks like TensorFlow, MXNet and Caffe.
H2O also has many technology partners, some like Angoss, IBM, RapidMiner and TIBCO Software also on this Magic Quadrant, which have integrated its technology platform and library.
This Zurich-based developer provides an open-source platform with commercial extensions for enterprise deployments.
Knime makes data science affordable through its commitment to open-source and well-priced commercial offerings. While the vendor doesn't put much into sales and marketing, it does focus on solutions engineering and partner programs.
"The vendor demonstrates a deep understanding of the market, a robust product strategy and strength across all use cases," Gartner said.
The Knime Analytics Platform took a step forward last year when it was introduced to Amazon Web Services and Microsoft Azure clouds. While there have been scalability complaints, the outlook to resolve them is improved by the recent cloud integrations.
This Boston-based company's analytics platform is well-rounded and easy to use. It can be leveraged by all sorts of data scientists.
The simplicity of the product-- an intuitive interface, easy access to data sources, simple programming for developing models and easily understood results -- explains RapidMiner's impressive revenue growth.
"RapidMiner continues to emphasize core data science and speed of model development and execution by introducing new productivity and performance capabilities," Gartner said.
But pricing often feels unpredictable. That's because the base of open-source components is being complemented by new capabilities that require a commercial license.
SAS, the industry leader in both total revenue and paying customers, offers many analytics products. But Gartner evaluated for this report only SAS Enterprise Miner (EM) and the SAS Visual Analytics suite
Gartner notes the Cary, N.C.-based vendor's "partner network enhances its visibility and support."
SAS has lost ground in the latest Magic Quadrant. That's partly due to a confusing multiproduct approach that's lessened the evaluation for Completeness of Vision, and to a perception of high licensing costs that's impaired its Ability to Execute.
SAS appeals to a broad variety of users, and excels in giving customers robust, enterprise-grade capabilities and high-quality support.
But as the market's focus shifts to open-source software and flexibility, SAS has lagged in creating a cohesive, open platform.
This Irvine, Calif.-based vendor offers a machine-learning platform that can be easily used by nonprofessional data scientists. Alteryx went public last year, putting it in a stronger position to invest in the capabilities of its platform, Alteryx Analytics.
Last year Alteryx was a Challenger, but it has advanced in Gartner's rankings because of its revenue and customer growth, customer satisfaction, "and a product vision focused on helping organizations instill a data and analytics culture without needing to hire expert data scientists."
Most customers still use the platform for data preparation, and Alteryx needs more aggressive marketing as a comprehensive data science platform. There's also room to improve enterprise functionality, including supporting Linux and running the platform on more than one server, according to Gartner.
Databricks offers a Spark-based platform as a cloud service. The vendor is new to Gartner's Data Science Magic Quadrant, but enjoys substantial traction in the market because it was founded by the creators of the open-source Spark project, making it the "Center of the Spark ecosystem."
The San Francisco-based company is leveraging that status to grow its customer base.
"Many companies introduce machine learning by using Spark as a starting point. Experienced organizations often select the Spark ecosystem to further strengthen their business," Gartner said.
Databricks sells proprietary features for security, reliability, operationalization, performance and real-time enablement on Amazon Web Services. A Microsoft Azure preview started in November 2017.
The company has been innovative in facilitating deployment onto cloud infrastructure, enabling machine-learning models to be run at large scale.
But it still needs to expand its market awareness.
Gartner evaluated IBM's SPSS (Modeler and Statistics) solutions. Data Science Experience (DSX), a different machine-learning offering, only factored into the Completeness of Vision assessment.
IBM remains a market-share leader. SPSS is a trusted and vetted enterprise solution with strengths in robust data preparation and the ability to operationalize and manage models.
"Its strategy, focused on the complete analytic pipeline, enables both expert and novice data scientists to be productive," Gartner said.
But IBM has lost ground from last year on both axes of the Magic Quadrant. A plan for a new interface that fully integrates SPSS Modeler into DSX has the potential to turn that around, with the DSX platform inspiring a more innovative vision.
But for now, IBM lacks a single, comprehensive, modern platform with SPSS Modeler undergoing renovation and DSX continuing to develop,
And customers rated poorly their experience using the platform.
While the tech giant sells many data science products, Gartner only evaluated Azure Machine Learning Studio. Other advanced analytics offerings influenced the assessment of Completeness of Vision.
Low scores for market responsiveness and product viability kept Microsoft from the Leaders Quadrant. Azure Machine Learning Studio's cloud-only nature limits its value for many users wanting to run on-premises.
But the cloud approach enables good flexibility, extensibility and openness. It also lends itself well to performance tuning and scalability. With frequent cloud software updates, users don't have to wait to take advantage of improvements.
Gartner noted Microsoft's commitment to open-source, deep-learning, streaming and Internet of Things use cases, and its focus on end-to-end analytic processes.
Microsoft is such a well-known brand that Azure Machine Learning has high visibility in the enterprise. But uptake has been slow, primarily due to the product's immaturity compared with rival offerings.
Domino Data Lab in San Francisco offers a comprehensive analytics platform geared for expert data scientists.
Domino's integration of open-source and proprietary tools earned it Gartner's highest overall score for flexibility, extensibility and openness.
Customers also praised the service and support they received from the company.
Domino provides excellent features to enable data scientists to offer transparency and work effectively with nontechnical users, Gartner said.
But holding Domino back is its complexity, which makes it unsuitable for amateur analysts. The platform also suffers in capability at the "beginning of the machine-learning life cycle" when data is accessed, ingested and prepared.
Dataiku's Data Science Studio focuses on enabling collaboration and maintaining enough simplicity to allow users to rapidly dive into machine-learning projects.
Customers told Gartner "the collaborative nature of Dataiku DSS has democratized machine learning across their organization."
The New York City-based company delivers an intuitive interface that makes its solution a popular choice for rapid prototyping.
But users also reported trouble deploying models in production environments, as well as problems with instabilities and bugs.
Lack of automation and higher-end prices also slow Dataiku's market adoption, noted Gartner.
Challenger: Tibco Software
The Palo Alto, Calif.-based company entered the machine-learning market, and this Magic Quadrant, with its acquisition of the Statistica platform from Quest Software last year. The platform, in only four years, had gone from Statsoft to Dell to Quest, with its vision changing several times along the way.
"It remains to be seen whether Statistica has finally found a permanent home that will support a long-term strategy," Gartner noted.
The business intelligence vendor's strength in the data science market was also boosted by the recent acquisition of Alpine Data, a solution that should soon be integrated with its larger ecosystem.
Tibco's Connected Intelligence strategy and accompanying products offer a strong ecosystem for Internet of Things use cases, noted Gartner.
But Statistica received one of the lowest overall scores from reference customers for performance and scalability. That could change as the platform integrates with other Tibco products.
The Natick, Mass.-based company is the developer of MATLAB, which has a large base of longstanding customers.
Despite the rapidly growing popularity of open-source technologies, MathWorks remains one of the most prominent vendors in the data science sector thanks to decades-long relationships with many customers.
But enterprise use cases for MATLAB are limited. The product favors engineering and science projects, and its vision doesn't extend well to marketing, sales or customer service.
MathWorks also is held back in the Magic Quadrant because of its reluctance to embrace open source.
Niche Player: SAP
The German software heavyweight has rebranded a platform with many components as SAP Predictive Analytics.
But SAP remains a Niche Player "due to low customer satisfaction scores, a lack of mind share, a fragmented toolchain, and significant technological weak spots (in relation to the cloud, deep learning, Python and notebooks, for example), relative to others," said Gartner.
SAP does have a unique vision of unifying machine learning across all its applications. And the new Predictive Analytics Integrator is a good start to those efforts. The platform is "especially good at automating many tasks and deploying across a range of business applications," Gartner said.
But customers complain of a poor experience using the platform, and an inability of SAP's products to meet their needs.
SAP lags in certain technological areas like cognitive computing and multi-cloud deployment.
Niche Player: Angoss
The Canadian company was acquired earlier this year by Datawatch.
Angoss has a good track record with banks, as well as in other sectors like insurance, transportation and utilities. Those customers are loyal, but the solution is still perceived as one for desktops, even though recent enterprise capabilities were added to bulk up the server environment.
Angoss is building on core products that are user-friendly and easy to use with support for open-source solutions like Spark ML, TensorFlow and H2O.ai. But after two decades in the industry, the company's market traction is weaker than would be expected.
Niche Player: Anaconda
Anaconda, based in Austin, Texas, was formerly known as Continuum Analytics.
Anaconda Enterprise 5.0 is an open-source development environment that implements an interactive-notebook concept. It integrates with other environments and open-source libraries.
Anaconda's strength lies in its ability to provide a central access point for Python developers looking to work together continuously on building machine-learning capabilities.
The platform is well-suited for professional data scientists comfortable with using the Python programming language. Nonprofessionals will struggle with it.
"The growing popularity of Python among data scientists gives Anaconda excellent visibility to developers," Gartner said, which is why there's a large and active community around Anaconda.
But the platform offers little in the way of automation.
The company is also struggling with converting open-source users to the enterprise product. Its sales efforts need to go beyond online marketing and evangelism.
Niche Player: Teradata
Teradata Unified Data Architecture (UDA) is an analytical ecosystem, geared for enterprises, that combines open-source and commercial technologies.
The San Diego-based company's platform includes Aster Analytics, a Teradata database, Hadoop and data management tools. But despite "strong operationalization capabilities," Gartner said, Teradata "still lacks a unified end-to-end technology platform."
Teradata excels at deploying machine-learning solutions at large scale. The company benefits from "a long history in industrial-strength data warehousing for multiple business sectors."
While performance and reliability rank high, the lack of cohesion and difficulty of using the platform relegate Teradata to the status of Niche Player.
The solution also faces pricing pressure as it requires "significant upfront investment."
The Article was originally published in CRN