In today’s hyper-fast cloud computing era, machine learning solutions drive exponential progress in improving systems. Machine learning’s ability to leverage Big Data analytics and identify patterns offers critical competitive advantage to today’s businesses.
Often used in combination with artificial intelligence and deep learning, machine learning uses sophisticated statistical modeling. These complex systems may reside in private cloud or public cloud. In any case, the passage of time boosts machine learning: as more data is added to a task and analyzed over time, ML produces more accurate the results.
The global machine learning market totaled $1.4 billion in 2017, according to BCC Research. It is estimated to top $8.8 billion by 2022, a jaw-dropping compound annual growth rate (CAGR) of 43.6 percent.
When considering which machine learning vendor to select, weigh these factors:
- Major platform or stand alone vendor: some of the ML vendors below offer a full cloud computing platform, others are stand alone firms dedicated to ML. There’s no hard and fast rule, but the stand alone firms may be more incentivized to bid competitively for your business.
- Type of machine learning: Do you want to implement ML for retail? For healthcare? Ask your potential vendor what sectors they have invested in the most heavily.
- Vendor’s future direction: Where directions, in terms of R&D and investment, will this vendor focus on in the years ahead? That answer to that question might help you find the best match for your long term goals.
The machine learning landscape is changing rapidly. Machine learning start-ups are constantly jumping into the space. Established vendors are introducing a variety of offerings that use machine learning in one form or another. Sorting through the choices and options can prove confusing. We’ve identified eight top vendors in the machine learning space, based on the features they offer, analyst reviews, customer testimonials and independent research.
- AWS SageMaker
- Google Machine Learning Engine
- IBM Watson Studio
- Microsoft Azure Machine Learning
- SAP Leonardo
- SAS Visual Data Mining and Machine Learning
Alteryx offers a unified platform that addresses data, analytics and machine learning. The vendor strives to minimize the need for data scientists by creating new and improved partnerships among IT, analytics experts and lines of business through a collaborative approach.
The product is available in both a code-free and code-friendly format. It uses a drag-and-drop workflow interface and ties into third party demographic, firmographic and spatial data. The company is rated as a “Leader” on the Gartner Magic Quadrant (MQ). Alteryx is moving toward powerful automation and rule-based recommendations, Gartner noted.
Alteryx offers integration with numerous major partners, including Tableau, AWS, Teradata, Microsoft, DataRobot, Salesforce, Oracle, Cloudera and Qlik. Machine learning functions feature parallel model analysis with predictive analytics, along with the ability to automate workflows and various processes.
The fully managed machine learning service allows data scientists and developers to build and train machine learning models and insert them into applications. This makes it possible to run parallel models, conduct flexible distributed training, and tackle A-B testing along with other tasks.
SageMaker is a fully managed AWS service that includes an integrated Jupyter authoring notebook instance. This aids in exploration and analysis by offering deep visibility into data sources. SageMaker offers a set of algorithms optimized for the environment and it supports proprietary algorithms and custom training scripts.
Sagemaker also features integration with Docker containers and Apache Spark. Together with AWS, it’s possible to use the integrated SageMaker console to launch and scale projects quickly. The platform supports HIPAA and other compliance standards.
Google Cloud Machine Learning (ML) Engine is a fully managed platform designed for developers and data scientists. The product allows users to construct high quality models using multiple machine learning frameworks, including scikit-learn, XGBoost, Keras, and TensorFlow.
The latter is a sophisticated deep learning framework that powers many Google products, from Google Photos to Google Cloud Speech. It allows users to deploy machine learning into production without the need for Docker containers. The highly scalable environment—which accommodates training on managed clusters, delivers a framework for automatically designing and evaluating model architectures. It is designed to accommodate existing workflows, and it scales to handle batch prediction and online prediction requirements.
Other key features include: automatic resource provisioning, the use of portable models, server-side processing and integrated cloud data storage.
IBM was a pioneer in the artificial intelligence (AI) and machine learning spaces. Today, Forrester ranks the company as a “Leader” among multimodal predictive analytics and machine learning solutions providers.
IBM Watson Studio is designed to accommodate a variety of independent platforms and various types of power users. This includes data engineers, app developers and data scientists. The result is strong collaboration capabilities. Among its top features: a robust architecture, strong algorithms and a powerful ability to execute machine learning.
The Watson Studio is part of an extensive collection of technologies and solutions that are available under the IBM Watson umbrella. It is designed to accommodate IBM’s SPSS modeling workflows but also plug in open source machine learning libraries and notebook-based interfaces. The solution is available in the cloud on or on the desktop. It incorporates visual drag-and-drop tools.
Azure Machine Learning Studio has emerged as a leading solution in the managed cloud space. It delivers a visual tool that aids developers, data scientists and non-data scientists in designing machine learning pipelines and solutions that address a wide variety of tasks. The platform offers a browser-based, visual drag-and-drop authoring environment that requires no coding. Gartner ranks Microsoft a “Visionary” in its MQ. The solution offers a high level of flexibly, extensibility and openness.
Azure ML Studio also delivers comprehensive capabilities across the full range of descriptive, diagnostic, predictive and prescriptive analytic types. Microsoft is continuing to expand the features and capabilities within Azure Machine Learning. This includes promoting deep learning through the Microsoft Cognitive Toolkit (formerly CNTK) as well as the joint ONNX open standard for neural networks. Microsoft also plans to automate a growing number of functions within the Azure Machine Learning platform.
RapidMiner offers data prep, machine learning and model deployment tools that deliver deep insight into data. It delivers a platform that is designed to tackle a variety of big data and analytics requirements.
In the machine learning space, the company has established itself as a heavyweight. It delivers a simple-to-use visual environment that incorporates automated model creation along with suggestions, guides and information about what approach might work best for a given situation. RapidMiner’s features and high level of flexibility have established it as a popular choice among machine learning solutions.
Data scientists and non-data scientists alike can easily inspect and customize projects and build on results. RapidMiner is ranked as a “Leader” in both the Gartner MQ and the Forrester Wave. It is among the highest rated data and machine learning solutions at Gartner Peer Insights.
Leonardo serves as an umbrella for a variety of SAP data products and services. It incorporates AI and machine learning through a cloud platform. This includes a machine learning foundation that allows developers, data scientists and others to create, run and deploy algorithms and machine learning apps.
The technology extracts data from text, images, natural language and other media to generate computer-based predictions. The highly scalable platform includes predefined and pretrained machine learning services, but it also incorporates TensorFlow models. This allows users to adopt a flexible framework to training and tuning machine learning. Forrester ranks SAP as a “strong performer” in the overall predictive analytics and machine learning category and cites the breadth and depth of service as a strong point.
However, Gartner rates the company a “Niche Player” in its MQ. It noted that SAP has made considerable strides in the data and machine learning spaces but still lags behind others in terms of features and capabilities.
A pioneer in the data mining and analytics spaces, SAS has established itself as a top choice among machine learning vendors. It offers Visual Data Mining and Machine Learning (VDMML), which ranked at the standout “Leader” in the Forrester Wave and as a top “Leader” in the Gartner MQ, though the research firm notes that the overall platform has lost some ground in terms of vision and ability to execute.
VDMML connects to numerous other products and data sources. Forrester noted that SAS has emerged as the first truly multimodal PAML solution. It uses a wizard to automatically train a model, convert it to code and plug in integrated features for automated machine learning. The unified environment also includes robust visual tools for creating models as well as deep neural networks.
In addition, data scientists can embed snippets of code directly into SAS programs to further enhance capabilities, including machine learning and deep learning output.
In addition to our top picks, there are other promising tools out there.
Splunk Enterprise provides a broad based platform that can be used for searching, monitoring, and analyzing data. The software can import data from a variety of sources, from logs to Big Data sources.
Data Science and Machine Learning Vendors at a Glance:
|Vendor and features||Alteryx||AWS SageMaker||Google Machine Learning Engine||IBM Watson Studio||Microsoft Azure Machine Learning||RapidMiner||SAP Leonardo||SAS Visual Data Mining and Machine Learning|
|ML Focus||Analytics and ML platform that allows users to build models in a single workflow||Hosted platform that supports supervised and unsupervised ML and deep learning.||Hosted platform that runs ML training jobs and predictions at scale.||Broad data science focus with cloud and desktop ML platforms.||Automated ML platform running on hosted Azure Cloud.||Highly automated ML platform suited for firms aiming to use machine learning broadly.||Broad and deep data science and ML platforms.||Multimodal PAML solution with powerful wizards to support data mining and ML. Strong automation.|
|Key Features and capabilities||Offers more than 250 code free pre-built tools and supports open source tools and scripting.||Zero-setup integrated Jupyter authoring notebook instance. Access to rich array of open source frameworks and tools.||Automatic resource provisioning and monitoring for CPUs, GPUs and TPUs. Strong open source support for tools and scripting languages.||Strong visual recognition and natural classification tools. Flexible approach that incorporates open source tools. Connects to IBM SPSS Modeler.||Strong descriptive, diagnostic, predictive and prescriptive analytics tools and capabilities. Strong support for open source ML tools and scripting.||Offers more than 1,500 machine learning and data prep functions, and it supports more than 40 files types. Connects to Amazon S3 and Dropbox.||Fully supports a BYOM approach. Strong open source integration. Runs in SAP’s HANA public cloud.||Includes best practice templates and in-memory processing that supports supervised and unsupervised learning.|
|User comments||High ratings for simplicity and usability. Complaints about the lack of robust reporting and visualization features.||High marks for the interface, configurability and usability. However, some say that the familiarity with AWS is crucial.||High ratings for functionality and integration.||Highly rated for features and capabilities. Some complaints revolving around using notebooks.||High ratings at Gartner Peer Review. Some complaints about support.||Among the highest rated data science and ML solutions. Users describe it as powerful and “revolutionary” though there are complaints about the lack of GPU support.||Customers give SAP high marks for usability and performance. Some complain about the lack of visualization tools for external data.||High ratings. Users like the powerful features and capabilities built into the platform, though some remark that there is a substantial learning curve associated with the product.|
|Pricing and licensing||Tiered system based on products used. Ranges from $5,195 to $33,800 per seat annually.||Complex tiered pricing model based on ML notebook instances and cloud resources consumed.||Based on resources and training hours used. Ranges from 1 cent per hour to $31 per hour for various services.||Tiered model from $99 per month per user to $6,000 per month per user or more at enterprise level.||Tiered pricing based on users and usage. Ranges from near zero to tens of thousands per user annually.||Tiered pricing ranging from $2,500 per user per year to upwards of $10,000 per user per year.||Tiered model based on node hours consumed in the cloud. Billing blocks are charge in increments of €7.25.||N/A|