Gartner’s 2021 Magic Quadrant cites ‘glut of innovation’ in information science and ML

Gartner’s Magic Quadrant report on data science and machine learning (DSLM) platform firms assesses what it says are the highest 20 distributors on this fast-growing trade section.

Knowledge scientists and different technical customers depend on these platforms to supply information, construct fashions, and use machine studying at a time when constructing machine studying functions is more and more changing into a approach for firms to distinguish themselves.

Gartner says AI remains to be “overhyped” however notes that the COVID-19 pandemic has made investments in DSLM extra sensible. Firms ought to deal with creating new use instances and functions for DSML — those which are seen and ship enterprise worth, Gartner mentioned within the report launched final week. Sensible firms ought to construct on profitable early initiatives and scale them.

The report evaluates DSML platforms’ scope, income and development, buyer counts, market traction, and product functionality scoring. Listed here are a number of the notable findings:

• Accountable AI governance, transparency, and addressing model-based biases are essentially the most beneficial differentiators on this market, and each listed vendor is making progress in these areas.
• Google and Amazon are lastly competing with Microsoft for supremacy when it comes to DSML capabilities within the cloud. Amazon wasn’t even included in final 12 months’s Magic Quadrant as a result of it hadn’t shipped its core product by the November 2019 cutoff date. The longest-standing huge names on this sector — IBM, MathWorks, and SAS — are nonetheless holding their floor and innovating with fashionable choices and adaptive methods.
• Quite a few smaller, youthful, and mid-size distributors are in sustained intervals of hypergrowth. The rising dimension of the market feeds startups in any respect phases of the info science lifecycle. Gartner observes that rising on the charge of the market really means rising slowly.
• Alibaba Cloud, Cloudera, and Samsung DDS are included within the Magic Quadrant for the primary time.
• The DSML platform software program market grew by 17.5% in 2019, producing $4 billion in income. It’s the second-fastest-growing section of the analytics and enterprise intelligence (BI) software program market behind fashionable BI platforms, which grew 17.9%. Its share of the general analytics and BI market grew to 16.1% in 2019.
• Essentially the most modern DSML distributors assist numerous varieties of customers collaborating on the identical venture: information engineers, professional information scientists, citizen information scientists, software builders, and machine studying specialists.

There stays a “glut of compelling improvements” and visionary roadmaps, Gartner says. That is an adolescent market, the place distributors are closely centered on innovation and differentiation, quite than pure execution. Gartner mentioned key areas of differentiation embrace UI, augmented DSML (AutoML), MLOps, efficiency and scalability, hybrid and multicloud assist, XAI, and cutting-edge use instances and strategies (reminiscent of deep studying, large-scale IoT, and reinforcement studying).

Above: Gartner Magic Quadrant for Knowledge Science and Machine Studying Platforms. (Supply: Gartner, March 2021)

Picture Credit score: Dataiku

Knowledge science and machine studying in 2021 and past

For many enterprises, the problem is to maintain up with the speedy tempo of change of their industries, pushed by how briskly their opponents, suppliers, and channel companions are digitally remodeling their companies.

• CIOs and senior administration groups need to perceive the specifics of how information science and machine studying fashions work. A high precedence for IT executives working with DSML applied sciences is knowing bias mitigation and the way DSML applied sciences can management for biases on a per-model foundation. Designing transparency ought to begin with mannequin and information repositories, offering larger visibility throughout a whole DSML platform.

• Enterprises proceed to wrestle with transferring extra AI fashions from pilot to manufacturing. In keeping with the 2020 Gartner AI in Organizations Survey, simply 53% of machine studying prototypes are ultimately deployed to manufacturing. Yield charges from the preliminary mannequin to manufacturing deployment present room for enchancment. Search for DSML distributors to step up their efforts to ship modeling apps and platforms that may settle for smaller datasets and nonetheless ship correct outcomes.
• Open supply software program (OSS) is a de facto customary with DSML distributors. OSS supplies enterprises the chance to get DSML initiatives up and working with little upfront spending. OSS adoption has turn out to be so pervasive that the majority DSML distributors depend on OSS, beginning with Python, essentially the most generally used language. DSML platform suppliers additionally assist optimize and curate OSS distributions.
• For any enterprise to put money into a DSML platform, integration and connectivity are important. DSML distributors are adopting parts for his or her platform architectures as a result of parts are extra extensible and could be tailor-made to an enterprise’s particular wants. Packaged fashions that combine right into a DSML platform utilizing APIs assist enterprises customise machine studying fashions for particular trade challenges they’re going through.
• Designing extra intuitive interfaces and workflows reduces the training curve for strains of enterprise and information analysts. Enhancements in augmented information science and ML assist offload all the info science and modeling work from skilled information scientists to enterprise analysts preferring to iterate fashions on their very own, usually altering constraints based mostly on market circumstances.
• Organizations depend on free and low-cost open supply, mixed with public cloud suppliers to scale back prices whereas experimenting with DSML initiatives. They’re then more likely to undertake industrial software program to sort out broader use instances and necessities for workforce collaboration and to maneuver fashions into manufacturing.

Which distributors are main — and why

Listed here are some company-specific insights included on this 12 months’s Magic Quadrant:

• SAS Visible Knowledge Mining and Machine Studying (VDMML) is the market chief, having dominated the Chief quadrant for years on this particular Magic Quadrant. Gartner provides SAS credit score for its cloud-native structure, automated function engineering and modeling, and area experience mirrored in its superior prototyping and manufacturing refinement use instances. SAS is usually seen as a legacy vendor that’s costly to implement and assist. The client loyalty SAS has accrued in world enterprises and the precedence its growth groups place on DSML helps the corporate preserve dominance on this market.
• IBM’s Watson Studio ascended into the Chief quadrant this 12 months, up from being thought of a Challenger in 2020. Gartner believes the corporate’s completeness of imaginative and prescient (horizontal axis of the quadrant) has improved since final 12 months, transferring it into the Chief quadrant. That is primarily as a result of IBM Watson Studio’s multi-persona assist, depth of accountable AI and governance, and part construction proving efficient for determination modeling. Constructing on a number of years of reinventing itself, IBM can ship an enterprise-class DSML that can efficiently progress past the pilot or proof-of-concept section. Gartner provides IBM credit score for capitalizing on earlier successes of SPSS, ILOG CPLEX Optimization Studio, earlier analytics merchandise, and the continuous stream of improvements from IBM Analysis.
• Alteryx’s robust momentum out there isn’t mirrored in its shift from the Chief quadrant to Challenger. Alteryx powered through last year’s uncertainty, reporting a 19% year-over-year enhance in income for 2020, reaching $495.3 million. Annual recurring income grew 32% 12 months over 12 months to achieve $492.6 million. Gartner provides Alteryx credit score for supporting a number of personas, a confirmed go-to-market technique, and delivering wonderful customer support and assist. Alteryx has confirmed to be modern, regardless of having that attribute talked about as a warning within the Magic Quadrant.
• Amazon SageMaker’s market momentum is formidable, additional strengthened by its tempo of innovation. In February, Amazon Internet Companies (AWS) introduced it has designed and can produce its personal machine learning training chip. AWS Trainium is designed to ship essentially the most teraflops of any machine studying coaching occasion within the cloud. AWS additionally introduced Trainium would assist all main frameworks (together with TensorFlow, PyTorch, and MXnet). Trainium will use the identical Neuron SDK utilized by AWS Inferentia (an AWS-designed chip for machine studying inference acceleration), making it straightforward for patrons to get began coaching shortly with AWS Trainium. AWS Trainium is coming to Amazon EC2 and Amazon SageMaker within the second half of 2021. Amazon SageMaker contains 12 parts: Studio, Autopilot, Floor Fact, JumpStart, Knowledge Wrangler, Characteristic Retailer, Make clear, Debugger, Mannequin Monitor, Distributed Coaching, Pipelines, and Edge Supervisor.
• Google will launch its unified AI Platform within the first quarter of 2021. That is after the cutoff date for analysis on this Magic Quadrant. It would launch key options like AutoML tables, XAI, AI platform pipelines, and different MLOps providers.

The challenges for DSML platform distributors in the present day start with balancing the wants for larger transparency and bias mitigation whereas creating and delivering modern new options at a predictable cadence. The Magic Quadrant displays present market actuality after updating with 4 new cloud distributors, one with an in depth ecosystem and confirmed market momentum.

One factor to contemplate after wanting on the Magic Quadrant is that there might be some mergers or acquisitions on the horizon. Search for BI distributors to both purchase or merge with DSML platform suppliers because the BI market’s route strikes towards augmented analytics and away from visualization. Additional fueling potential M&A exercise is the truth that DSML platforms may use enhanced information transformation and discovery assist on the mannequin stage, which is a long-standing power of BI platforms.