Databricks

The Year in Machine Learning (Part Four)

This is the fourth installment in a four-part review of 2016 in machine learning and deep learning.

— Part One covered Top Trends in the field, including concerns about bias, interpretability, deep learning’s explosive growth, the democratization of supercomputing, and the emergence of cloud machine learning platforms.

— Part Two surveyed significant developments in Open Source machine learning projects, such as R, Python, Spark, Flink, H2O, TensorFlow, and others.

— Part Three reviewed the machine learning and deep learning initiatives of Big Tech Brands, industry leaders with significant budgets for software development and marketing.

In Part Four, I profile eleven startups in the machine learning and deep learning space. A search for “machine learning” in Crunchbase yields 2,264 companies. This includes companies, such as MemSQL, who offer absolutely no machine learning capability but hype it anyway because Marketing; it also includes application software and service providers, such as Zebra Medical Imaging, who build machine learning into the services they provide.

All of the companies profiled in this post provide machine learning tools as software or services for data scientists or for business users. Within that broad definition, the firms are highly diverse:

— Continuum Analytics, Databricks, and H2O.ai drive open source projects (Anaconda, Apache Spark, and H2O, respectively) and deliver commercial support.

— Alpine Data, Dataiku, and Domino Data Lab offer commercially licensed collaboration tools for data science teams. All three run on top of an open source platform.

— KNIME and RapidMiner originated in Europe, where they have large user communities. Both combine a business user interface with the ability to work with Big Data platforms.

— Fuzzy Logix and Skytree provide specialized capabilities primarily for data scientists.

— DataRobot delivers a fully automated workflow for predictive analytics that appeals to data scientists and business users. It runs on an open source platform.

Four companies deserve an “honorable mention” but I haven’t profiled them in depth:

— Two startups, BigML and SkyMind, are still in seed funding stage. I don’t profile them below, but they are worth watching. BigML is a cloud-based machine learning service; SkyMind drives the DL4J open source project for deep learning.

— Two additional companies aren’t startups because they’ve been in business for more than thirty years. Salford Systems developed the original software for CART and Random Forests; the company has added more techniques to its suite over time and has a loyal following. Statistica, recently jettisoned by Dell, delivers a statistical package with broad capabilities; the company consistently performs well in user satisfaction surveys.

I’d like to take a moment to thank those who contributed tips and ideas for this series, including Sri Ambati, Betty Candel, Leslie Miller, Bob Muenchen, Thomas Ott, Peter Prettenhofer, Jesus Puente, Dan Putler, David Smith, and Oliver Vagner.

Alpine Data

In 2016, the company formerly known as Alpine Data Labs changed its name and CEO. Alpine dropped the “Labs” from its brand — I guess they didn’t want to be confused with companies that test stool samples — so now it’s just Alpine Data. And, ex-CEO Joe Otto is now an “Advisor,” replaced by Dan Udoutch, a “seasoned executive” with 30+ years of experience in business and zero years of experience in machine learning or advanced analytics. The company also dropped its CFO and head of Sales during the year, presumably because the investors were extremely happy with Alpine’s business results.

Originally built to run in Greenplum database, the company ported some of its algorithms to MapReduce in early 2013. Riding a wave of Hadoop buzz, Alpine closed on a venture round in November 2013, just in time for everyone to realize that MapReduce sucks for machine learning. The company quickly turned to Spark — Databricks certified Alpine on Spark in 2014 — and has gradually ported its analytics operators to the new framework.

It seems that rebuilding on Spark has been a bit of a slog because Alpine hasn’t raised a fresh round of capital since 2013. As a general rule, startups that make their numbers get fresh rounds every 12-24 months; companies that don’t get fresh funding likely aren’t making their numbers. Investors aren’t stupid and, like the dog that did not bark, a venture capital round that does not happen says a lot about a company’s prospects.

In product news, the company announced Chorus 6, a major release, in May, and Chorus 6.1 in September. Enhancements in the new releases include:

— Integration with Jupyter notebooks.

— Additional machine learning operators.

— Spark auto-tuning. Chorus pushes processing to Spark, and Alpine has developed an optimizer to tune the generated Spark code.

— PFA support for model export. This is excellent, a cutting edge feature.

— Runtime performance improvements.

— Tweaks to the user experience.

Lawrence Spracklen, Alpine’s VP of Engineering, will speak about Spark auto-tuning at the Spark Summit East in Boston.

Prospective users and customers should look for evidence that Alpine is a viable company, such as a new funding round, or audited financials that show positive cash flow.

Continuum Analytics

Continuum Analytics develops and supports Anaconda, an open source Python distribution for data science. The core Anaconda bundle includes Navigator, a desktop GUI that manages applications, packages, environments and channels; 150 Python packages that are widely used in data science; and performance optimizations. Continuum also offers commercially licensed extensions to Anaconda for scalability, high performance and ease of use.

Anaconda 2.5, announced in February, introduced performance optimization with the Intel® Math Kernel Library. Beginning with this release, Continuum bundled Anaconda with Microsoft R Open, an enhanced free R distribution.

In 2016, Continuum introduced two major additions to the Anaconda platform:

— Anaconda Enterprise Notebooks, an enhanced version of Jupyter notebooks

— Anaconda Mosaic, a tool for cataloging heterogeneous data

The company also announced partnerships with Cloudera, Intel, and IBM. In September, Continuum disclosed $4 million in equity financing. The company was surprisingly quiet about the round — there was no press release — possibly because it was undersubscribed.

Continuum’s AnacondaCon 2017 conference meets in Austin February 7-9.

Databricks leads the development of Apache Spark (profiled in Part Two of this review) and offers a cloud-based managed service built on Spark. The company also offers training, certification, and organizes the Spark Summits.

The team that originally developed Spark founded Databricks in 2013. Company employees continue to play a key role in Apache Spark, holding a plurality of the seats on the Project Management Committee and contributing more new code to the project than any other company.

In 2016, Databricks added a dashboarding tool and a RESTful interface for job and cluster management to its core managed service. The company made major enhancements to the Databricks security framework, completed SOC 2 Type 1 certification for enterprise security, announced HIPAA compliance and availability in Amazon Web Services’ GovCloud for sensitive data and regulated workloads.

Databricks also launched a free Community edition; a five-part series of free MOOCs; completed its annual survey of the Spark user community, and organized three Spark Summits.

In December, Databricks announced a $60 million “C” round of venture capital. New Enterprise Associates led the round; Andreessen Horowitz participated.

Dataiku

Dataiku develops and markets Data Science Studio (DSS), a workflow and collaboration environment for machine learning and advanced analytics. Users interact with the software through a drag-and-drop interface; DSS pushes processing down to Hadoop and Spark. The product includes connectors to a wide variety of file systems, SQL platforms, cloud data stores and NoSQL databases.

In 2016, Dataiku delivered Releases 3.0 and 3.1. Major new capabilities include H2O integration (through Sparkling Water); additional data sources (IBM Netezza, SAP HANA, Google BigQuery, and Microsoft Azure Data Warehouse); added support for Spark MLLib algorithms; performance improvements, and many other enhancements.

In October, Dataiku closed on a $14 million “A” round of venture capital. FirstMark Capital led the financing, with participation from Serena Capital.

DataRobot

DataRobot, a Boston-based startup founded by insurance industry veterans, offers an automated machine learning platform that combines built-in expertise with a test-and-learn approach. Leveraging an open source back end, the company’s eponymous software searches through combinations of algorithms, pre-processing steps, features, transformations and tuning parameters to identify the best model for a particular problem.

The company has a team of Kaggle-winning data scientists and leverages this expertise to identify new machine learning algorithms, feature engineering techniques, and optimization methods. In 2016, DataRobot added several new capabilities to its product, including support for Hadoop deployment, deep learning with TensorFlow, reason codes that explain prediction, feature impact analysis, and additional capabilities for model deployment.

DataRobot also announced major alliances with Alteryx and Cloudera. Cloudera awarded the company its top-level certification: the software integrates with Spark, YARN, Cloudera Service Descriptors, and Cloudera Parcels.

Earlier in the year, DataRobot closed on $33 million in Series B financing. New Enterprise Associates led the round; Accomplice, Intel Capital, IA Ventures, Recruit Strategic Partners, and New York Life also participated.

Domino Data Lab

Domino Data Lab offers the Domino Data Science Platform (DDSP) a scalable collaboration environment that runs on-premises, in virtual private clouds or hosted on Domino’s AWS infrastructure.

DDSP provides data scientists with a shared environment for managing projects, scalable computing with a variety of open source and commercially licensed software, job scheduling and tracking, and publication through Shiny and Flask. Domino supports rollbacks, revision history, version control, and reproducibility.

In November, Domino announced that it closed a $10.5 million “A” round led by Sequoia Capital. Bloomberg Beta, In-Q-Tel, and Zetta Venture Partners also participated.

Fuzzy Logix

Fuzzy Logix markets DB Lytix, a library of more than eight hundred functions for machine learning and advanced analytics. Functions run as database table functions in relational databases (Informix, MySQL, Netezza, ParAccel, SQL Server, Sybase IQ, Teradata Aster and Teradata Database) and in Hadoop through Hive.

Users invoke DB Lytix functions from SQL, R, through BI tools or from custom web interfaces. Functions support a broad range of machine learning capabilities, including feature engineering, model training with a rich mix of supported algorithms, plus simulation and Monte Carlo analysis. All functions support native in-database scoring. The software is highly extensible, and Fuzzy Logix offers a team of well-qualified consultants and developers for custom applications.

In April, the company announced the availability of DB Lytix on Teradata Aster Analytics, a development that excited all three of the people who think Aster has legs.

H2O.ai

H2O.ai develops and supports H2O, the open source machine learning project I profiled in Part Two of this review. As I noted in Part Two, H2O.ai updated Sparkling Water, its Spark integration for Spark 2.0; released Steam, a model deployment framework, to production, and previewed Deep Water, an interface to GPU-accelerated back ends for deep learning.

In 2016, H2O.ai added 3,200 enterprise organizations and over 43,000 users to its roster, bringing its open source community to over 8,000 enterprises and nearly 70,000 users worldwide. In the annual KDnuggets poll of data scientists, reported usage tripled. New customers include Kaiser Permanente, Progressive, Comcast, HCA, McKesson, Macy’s, and eBay.

KNIME

KNIME.com AG, a commercial enterprise based in Zurich, Switzerland, distributes the KNIME Analytics Platform under a GPL license with an exception permitting third parties to use the API for proprietary extensions. The KNIME Analytics Platform features a graphical user interface with a workflow metaphor. Users build pipelines of tasks with drag-and-drop tools and run them interactively or in batch.

KNIME offers commercially licensed extensions for scalability, integration with data platforms, collaboration, and productivity. The company provides technical support for the extension software.

During the year, KNIME delivered two dot releases and three maintenance releases. The new features added to the open source edition in Releases 3.2 and 3.3 include Workflow Coach, a recommender based on community usage statistics; streaming execution; feature selection; ensembles of trees and gradient boosted trees; deep learning with DL4J, and many other enhancements. In June, KNIME launched the KNIME Cloud Analytics Platform on Microsoft Azure.

KNIME held its first Summit in the United States in September and announced the availability of an online training course available through O’Reilly Media.

RapidMiner

RapidMiner, Inc. of Cambridge, Massachusetts, develops and supports RapidMiner, an easy-to-use package for business analysis, predictive analytics, and optimization. The company launched in 2006 (under the corporate name of Rapid-I) to drive development, support, and distribution for the RapidMiner software project. The company moved its headquarters to the United States in 2013.

The desktop version of the software, branded as RapidMiner Studio, is available in free and commercially licensed editions. RapidMiner also offers a commercially licensed Server edition, and Radoop, an extension that pushes processing down to Hive, Pig, Spark, and H2O.

RapidMiner introduced Release 7.x in 2016 with an updated user interface. Other enhancements in Releases 7.0 through 7.3 include a new data import facility, Tableau integration, parallel cross-validation, and H2O integration (featuring deep learning, gradient boosted trees and generalized linear models).

The company also introduced a feature called Single Process Pushdown. This capability enables RapidMiner users to supplement native Spark and H2O algorithms with RapidMiner pipelines for execution in Hadoop. RapidMiner supports Spark 2.0 as of Release 7.3.

In January 2016, RapidMiner closed a $16 million equity round led by Nokia Growth Partners. Ascent Venture Partners, Earlybird Venture Capital, Longworth Venture Partners, and OpenOcean also participated.

Skytree

Skytree Inc. develops and markets an eponymous commercially licensed software package for machine learning. Its founders launched the venture in 2012 to monetize an academic machine learning project (Georgia Tech’s FastLab).

The company landed an $18 million venture capital round in 2013 and hasn’t secured any new funding since then. (Read my comments under Alpine Data to see what that indicates.) Moreover, the underlying set of algorithms does not seem to have changed much since then, though Skytree has added and dropped several different add-ons and wrappers.

Users interact with the software through the Skytree Command Line Interface (CLI), Java and Python APIs or a browser-based GUI. Output includes explanations of the model in plain English. Skytree has a grid search feature for parameterization, which it trademarks as AutoModel, labels as “ground-breaking” and is attempting to patent. Analysts who don’t know anything about grid search think this is amazing.

In 2016, Skytree introduced a freemium edition, branded as Skytree Express. Hold out another six months and they’ll pay you to try it.

As is the case with Alpine Data, if you like Skytree’s technology wait for another funding round, or ask the company to provide evidence of positive cash flow.

The Year in Machine Learning (Part Two)

This is the second installment in a four-part review of 2016 in machine learning and deep learning. Part One, here, covered general trends. In Part Two, we review the year in open source machine learning and deep learning projects. Parts Three and Four will cover commercial machine learning and deep learning software and services.

There are thousands of open source projects on the market today, and we cannot cover them all. We’ve selected the most relevant projects based on usage reported in surveys of data scientists, as well as development activity recorded in OpenHub. In this post, we limit the scope to projects with a non-profit governance structure, and those offered by commercial ventures that do not also provide licensed software. Part Three will include software vendors who offer open source “community” editions together with commercially licensed software.

R and Python maintained their leadership as primary tools for open data science. The Python versus R debate continued amid an emerging consensus that data scientists should consider learning both. R has a stronger library of statistics and machine learning techniques and is agiler when working with small data. Python is better suited to developing applications, and the Python open source license is less restrictive for commercial application development.

Not surprisingly, deep learning frameworks were the most dynamic category, with TensorFlow, Microsoft Cognitive, and MXNet taking leadership away from more mature tools like Caffe and Torch. It’s remarkable that deep learning tools introduced as recently as 2014 now seem long in the tooth.

The R Project

The R user community continued to expand in 2016. It ranked second only to SQL in the 2016 O’Reilly Data Science Salary Survey; first in the KDNuggets poll; and first in the Rexer survey. R ranked fifth in the IEEE Spectrum ranking.

R functionality grew at a rapid pace. In April, Microsoft’s Andrie de Vries reported that there were more than 8,000 packages in CRAN, R’s primary repository for contributed packages. As of mid-December, there are 9,737 packages. Machine learning packages in CRAN continued to grow in number and functionality.

The R Consortium, a Collaborative Project of the Linux Foundation, made some progress in 2016. IBM and ESRI joined the Consortium, whose membership now also includes Alteryx, Avant, DataCamp, Google, Ketchum Trading, Mango Solutions, Microsoft, Oracle, RStudio, and TIBCO. There are now three working groups and eight funded projects.

Hadley Wickham had a good year. One of the top contributors to the R project, Wickham co-wrote R for Data Science and released tidyverse 1.0.0 in September. In The tidy tools manifesto, Wickham explained the four basic principles to a tidy API.

Max Kuhn, the author of Applied Predictive Modeling and developer of the caret package for machine learning, joined RStudio in November. RStudio previously hired Joseph Rickert away from Microsoft.

AT&T Labs is doing some impressive work with R, including the development of a distributed back-end for out-of-core processing with Hadoop and other data platforms. At the UseR! Conference, Simon Urbanek presented a summary.

It is impossible to enumerate all of the interesting analysis performed in R this year. David Robinson’s analysis of Donald Trump’s tweets resonated; using tidyverse, tidytext, and twitteR, Robinson was able to distinguish between the candidate’s “voice” and that of his staffers on the same account.

On the Revolutions blog, Microsoft’s David Smith surveyed the growing role of women in the R community.

Microsoft and Oracle continued to support enhanced R distributions; we’ll cover these in Part Three of this survey.

Python

Among data scientists surveyed in the 2016 KDNuggets poll, 46% said they use Python for analytics, data mining, data science or machine learning projects in the past twelve months. That figure was up from 30% in 2015, and second only to R. In the 2016 O’Reilly Data Science Salary Survey, Python ranked third behind SQL and R.

Python Software Foundation (PSF) expanded the number and dollar value of its grants. PSF awarded many small grants to groups around the world that promote Python education and training. Other larger grants went to projects such as the design of the Python in Education site, improvements to the packaging ecosystem (see below), support for the Python 3.6 beta 1 release sprint, and support for major Python conferences.

The Python Packaging Authority launched the Warehouse project to replace the existing Python Packaging Index (PyPI.) Goals of the project include updating the visual identity, making packages more discoverable and improving support for package users and maintainers.

PSF released Python 3.6.0 and Python 2.7.13 in December. The scikit-learn team released Version 0.18 with many enhancements and bug fixes; maintenance release Version 0.18.1 followed soon after that.

Many of the key developments for machine learning in Python were in the form of Python APIs to external packages, such as Spark, TensorFlow, H2O, and Theano. We cover these separately below.

Continuum Analytics expanded its commercial support for Python during the year and added commercially licensed software extensions which we will cover in Part Three.

Apache Software Foundation

There are ten Apache projects with machine learning capabilities. Of these, Spark has the most users, active contributors, commits, and lines of code added. Flink is a close second in active development, although most Flink devotees care more about its event-based streaming than its machine learning capabilities.

Top-Level Projects

There are four top-level Apache projects with machine learning functionality: Spark, Flink, Mahout, and OpenNLP.

Apache Spark

The Spark team delivered Spark 2.0, a major release, and six maintenance releases. Key enhancements to Spark’s machine learning capabilities in this release included additional algorithms in the DataFrames-based API, in PySpark and in SparkR, as well as support for saving and loading ML models and pipelines. The DataFrames-based API is now the primary interface for machine learning in Spark, although the team will continue to support the RDD-based API.

GraphX, Spark’s graph engine, remained static. Spark 2.0 included many other enhancements to Spark’s SQL and Streaming capabilities.

Third parties added 24 machine learning packages to Spark Packages in 2016.

The Spark user community continued to expand. Databricks reported 30% growth in Spark Summit attendees and 240% growth in Spark Meetup members. 18% of respondents to Databricks’ annual user survey reported using Spark’s machine learning library in production, up from 13% in 2015. Among data scientists surveyed in the 2016 KDNuggets poll, 22% said they use Spark; in the 2016 O’Reilly Data Science Salary Survey, 21% of the respondents reported using Spark.

The Databricks survey also showed that 61% of users work with Spark in the public cloud, up from 51% in 2015. As of December 2016, there are Spark services available from each of the major public cloud providers (AWS, Microsoft, IBM and Google), plus value-added managed services for data scientists from Databricks, Qubole, Altiscale and Domino Data.

Apache Flink

dataArtisans’ Mike Winters reviewed Flink’s accomplishments in 2016 without using the words “machine learning.” That’s because Flink’s ML library is still pretty limited, no doubt because Flink’s streaming runtime is the primary user attraction.

While there are many use cases for scoring data streams with predictive models, there are few real-world use cases for training predictive models on data streams. Machine learning models are useful when they generalize to a population, which is only possible when the process that creates the data is in a steady state. If a process is in a steady state, it makes no difference whether you train on batched data or streaming data; the latest event falls into the same mathematical space as previous events. If recent events produce major changes to the model, the process is not in a steady state, so we can’t rely on the model to predict future events.

Flink does not yet support PMML model import, a relatively straightforward enhancement that would enable users to generate predictions on streaming data with models built elsewhere. Most streaming engines support this capability.

There may be use cases where Flink’s event-based streaming is superior to Spark’s micro-batching. For the most part, though, Flink strikes me as an elegant solution looking for a problem to solve.

Apache Mahout

The Mahout team released four double-dot releases. Key enhancements include the Samsara math environment and support for Flink as a back end. Most of the single machine and MapReduce algorithms are deprecated, so what’s left is a library of matrix operators for Spark, H2O, and Flink.

Apache OpenNLP

OpenNLP is a machine learning toolkit for processing natural language text. It’s not dead; it’s just resting.

Incubator Projects

In 2016, two machine learning projects entered the Apache Incubator, while no projects graduated, leaving six in process at the end of the year: SystemML, PredictionIO, MADLib, SINGA, Hivemall, and SAMOA. SystemML and Hivemall are the best bets to graduate in 2017.

Apache SystemML

SystemML is a library of machine learning algorithms that run on Spark and MapReduce, originally developed by IBM Research beginning in 2010. IBM donated the code to Apache in 2015; since then, IBM has committed resources to developing the project. All of the major contributors are IBM employees, which begs the question: what is the point of open-sourcing software if you don’t attract a community of contributors?

The team delivered three releases in 2016, adding algorithms and other features, including deep learning and GPU support. Given the support from IBM, it seems likely that the project will hit Release 1.0 this year and graduate to top-level status.

Usage remains light among people not employed by IBM. There is no “Powered By SystemML” page, which implies that nobody else uses it. IBM added SystemML to BigInsights this year, which expands the potential reach to IBM-loyal enterprises if there are any of those left. It’s possible that IBM uses the software in some of its other products.

Apache PredictionIO

PredictionIO is a machine learning server built on top of an open source stack, including Spark, HBase, Spray, and Elasticsearch. An eponymous startup began work on the project in 2013; Salesforce acquired the company earlier this year and donated the assets to Apache. Apache PredictionIO entered the Apache Incubator in May.

Apache PredictionIO includes many templates for “prebuilt” applications that use machine learning. These include an assortment of recommenders, lead scoring, churn prediction, electric load forecasting, sentiment analysis, and many others.

Since entering the Incubator, the team has delivered several minor releases. Development activity is light, however, which suggests that Salesforce isn’t doing much with this.

Apache SINGA

SINGA is a distributed deep learning project originally developed at the National University of Singapore and donated to Apache in 2015. The platform currently supports feed-forward models, convolutional neural networks, restricted Boltzmann machines, and recurrent neural networks. It includes a stochastic gradient descent algorithm for model training.

The team has delivered three versions in 2016, culminating with Release 1.0.0 in September. The release number suggests that the team thinks the project will soon graduate to top-level status; they’d better catch up with paperwork, however, since they haven’t filed status reports with Apache in eighteen months.

Apache MADLib

MADLib is a library of machine learning functions that run in PostgreSQL, Greenplum Database and Apache HAWQ (incubating). Work began in 2010 as a collaboration between researchers at UC-Berkeley and data scientists at EMC Greenplum (now Pivotal Software). Pivotal donated the software assets to the Apache Software Foundation in 2015, and the project entered Apache incubator status.

In 2016, the team delivered three minor releases. The active contributor base is tiny, averaging three contributors per month.

According to a survey conducted by the team, most users have deployed the software on Greenplum database. Since Greenplum currently ranks 35th in the DB-Engines popularity ranking and is sinking fast, this project doesn’t have anywhere to go unless the team can port it to a broader set of platforms.

Apache Hivemall

Originally developed by Treasure Data and donated to the Apache Software Foundation, Hivemall is a scalable machine learning library implemented as a collection of Hive UDFs designed to run on Hive, Pig or Spark SQL with MapReduce, Tez or Spark. The team organized in September 2016 and plans an initial release in Q1 2017.

Given the relatively mature state of the code, large installed base for Hive, and high representation of Spark committers on the PMC, Hivemall is a good bet for top-level status in 2017.

Apache SAMOA

SAMOA entered the Apache Incubator two years ago and died. It’s a set of distributed streaming machine learning algorithms that run on top of S4, Storm, and Samza.

As noted above, under Flink, there isn’t much demand for streaming machine learning. S4 is moribund, Storm is old news and Samza is going nowhere; so, you can think of SAMOA as like an Estate Wagon built on an Edsel chassis. Unless the project team wants to port the code to Spark or Flink, this project is toast.

Machine Learning Projects

This category includes general-purpose machine learning platforms that support an assortment of algorithms for classification, regression, clustering and association. Based on reported usage and development activity, we cover H2O, XGBoost, and Weka in this category.

Three additional projects are worth noting, as they offer graphical user interfaces and appeal to business users. KNIME and RapidMiner provide open-source editions of their software together with commercially licensed versions; we cover these in Part Three of this survey. Orange is a project of the Bioinformatics Laboratory, Faculty of Computer and Information Science, University of Ljubljana, Slovenia.

Vowpal Wabbit gets an honorable mention. Known to Kaggleists as a fast and efficient learner, VW’s user base is currently too small to warrant full coverage. The project is now domiciled at Microsoft Research. It will be interesting to see if MSFT does anything with it.

H2O

H2O is an open source machine learning project of H2O.ai, a commercial venture. (We’ll cover H2O.ai’s business accomplishments in Part Three of this report.)

In 2016, the H2O team updated Sparkling Water for compatibility with Spark 2.0. Sparkling Water enables data scientists to combine Spark’s data ingestion and ETL capabilities with H2O machine learning algorithms. The team also delivered the first release of Steam, a component that supports model management and deployment at scale, and a preview of Deep Water for deep learning.

For 2017, H2O.ai plans to add an automated machine learning capability and deliver a production release of Deep Water, with support for TensorFlow, MXNet and Caffe back ends.

According to H2O.ai, H2O more than doubled its user base in 2016.

XGBoost

A project of the University of Washington’s Distributed Machine Learning Common (DMLC), XGBoost is an optimized distributed gradient boosting library used by top data scientists, who appreciate its scalability and accuracy. Tianqi Chen and Carlos Guestrin published a paper earlier this year describing the algorithm. Machine learning startups DataRobot and Dataiku added XGBoost to their platforms in 2016.

Weka

Weka is a collection of machine learning algorithms written in Java, developed at the University of Waikato in New Zealand and distributed under GPU license. Pentaho and RapidMiner include the software in their commercial products.

We include Weka in this review because it is still used by a significant minority of data scientists; 11% of those surveyed in the annual KDnuggets poll said they use the software. However, reported usage is declining rapidly, and development has virtually flatlined in the past few years, which suggests that this project may go the way of the eponymous flightless bird.

Deep Learning Frameworks

We include in this category software whose primary purpose is deep learning. Many general-purpose machine learning packages also support deep learning, but the packages listed here are purpose-built for the task.

Since they were introduced in late 2015, Google’s TensorFlow and Microsoft’s Cognitive Toolkit have rocketed from nothing to leadership in the category. With backing from Amazon and others, MXNet is coming on strong, while Theano and Keras have active communities in the Python world. Meanwhile, older and more mature frameworks, such as Caffe, DL4J, and Torch, are getting buried by the new kids on the block.

Money talks; commercial support matters. It’s a safe bet that projects backed by Google, Microsoft and Amazon will pull away from the pack in 2017.

TensorFlow

TensorFlow is the leading deep learning framework, measured by reported usage or by development activity. Launched in 2015, Google’s deep learning platform went from zero to leadership in record time.

In April, Google released TensorFlow 0.8, with support for distributed processing. The development team shipped four additional releases during the year, with many additional enhancements, including:

Python 3.5 support
iOS support
Microsoft Windows support (selected functions)
CUDA 8 support
HDFS support
k-Means clustering
WALS matrix factorization
Iterative solvers for linear equations, linear least squares, eigenvalues and singular values

Also in April, DeepMind, Google’s AI research group, announced plans to switch from Torch to TensorFlow.

Google released its image captioning model in TensorFlow in September. The Google Brain team reported that this model correctly identified 94% of the images in the ImageNet 2012 benchmark.

In December, Constellation Research selected TensorFlow as 2016’s best innovation in enterprise software, citing its extensive use in projects throughout Google and strong developer community.

Microsoft Cognitive Toolkit

In 2016, Microsoft rebranded its deep learning framework as Microsoft Cognitive Toolkit (MCT) and released Version 2.0 to beta, with a new Python API and many other enhancements. In VentureBeat, Jordan Novet reports.

At the Neural Information Processing Systems (NIPS) Conference in early December, Cray announced that it successfully ran MCT on a Cray XC50 supercomputer with more than 1,000 NVIDIA Tesla P100 GPU accelerators.

Separately, Microsoft and NVIDIA announced a collaborative effort to support MCT on Tesla GPUs in Azure or on-premises, and on the NVIDIA DGX-1 supercomputer with Pascal GPUs.

Theano

Theano, a project of the Montreal Institute for Learning Algorithms at the University of Montreal, is a Python library for computationally intensive scientific investigation. It allows users to efficiently define, optimize and evaluate mathematical expressions with multi-dimensional arrays. (Reference here.) Like CNTK and TensorFlow, Theano represents neural networks as a symbolic graph.

The team released Theano 0.8 in March, with support for multiple GPUs. Two additional double-dot releases during the year added support for CuDNN v.5 and fixed bugs.

MXNet

MXNet, a scalable deep learning library, is another project of the University of Washington’s Distributed Machine Learning Common (DMLC). It runs on CPUs, GPUs, clusters, desktops and mobile phones, and supports APIs for Python, R, Scala, Julia, Matlab, and Javascript.

The big news for MXNet in 2016 was its selection by Amazon Web Services. Craig Matsumoto reports; Serdar Yegulalp explains; Eric David dives deeper; Martin Heller reviews.

Keras

Keras is a high-level neural networks library that runs on TensorFlow or Theano. Originally authored by Google’s Francois Chollet, Keras had more than 200 active contributors in 2016.

In the Huffington Post, Chollet explains how Keras differs from other DL frameworks. Short version: Keras abstracts deep learning architecture from the computational back end, which made it easy to port from Theano to TensorFlow.

DL4J

Updated, based on comments from Skymind CEO Chris Nicholson.

Deeplearning4j (DL4J) is a project of Skymind, a commercial venture. IT is an open-source, distributed deep-learning library written for Java and Scala. Integrated with Hadoop and Spark, DL4J runs on distributed GPUs and CPUs. Skymind benchmarks well against Caffe, TensorFlow, and Torch.

While Amazon, Google, and Microsoft promote deep learning on their cloud platforms, Skymind seeks to deliver deep learning on standard enterprise architecture, for organizations that want to train models on premises. I’m skeptical that’s a winning strategy, but it’s a credible strategy. Skymind landed a generous seed round in September, which should keep the lights on long enough to find out. Intel will like a deep learning framework that runs on Xeon boxes, so there’s a possible exit.

Skymind proposes to use Keras for a Python API, which will make the project more accessible to data scientists.

Caffe

Caffe, a project of the Berkeley Vision and Learning Center (BVLC) is a deep learning framework released under an open source BSD license. Stemming from BVLC’s work in vision and image recognition, Caffe’s core strength is its ability to model a Convolutional Neural Network (CNN). Caffe is written in C++. Users interact with Caffe through a Python API or through a command line interface. Deep learning models trained in Caffe can be compiled for operation on most devices, including Windows.

I don’t see any significant news for Caffe in 2016.

Databricks Releases Spark Survey

In a press release and blog post, Databricks announces results from its 2016 Spark Survey. Databricks surveyed 1,615 Spark users and prospective users in July, 2016 Respondents include data engineers, data scientists, architects, technical managers, and academics.

Key findings from the survey:

Spark SQL remains the most widely used component.
- 88% use Spark SQL
- 71% use Spark Streaming
- 71% use MLlib (machine learning)
Respondents value Spark’s performance and advanced analytics.
- 91% rate performance very important
- 82% rate advanced analytics very important
- 76% rate ease of programming very important
- 69% rate ease of deployment very important
- 51% rate real-time streaming very important
Production use has increased markedly since 2015.
- 40% use SQL in production, up from 24%
- 38% use DataFrames in production, up from 15%
- 22% use streaming in production, up from 14%
- 18% use machine learning, up from 13%
So has usage in the public cloud.
- 61% said they use Spark in the public cloud, up from 51% in 2015.
Usage of Spark deployed on-premises has declined.
- 42% use Spark in a standalone deployment, down from 48%
- 36% use Spark under YARN, down from 40%
- 7% use Spark on Apache Mesos, down from 11%
The Scala API remains the most popular, followed closely by the Python API.
- 65% use Scala, down from 71% in 2015
- 62% use Python, up from 58%
- 44% use SQL, up from 36%
- 29% use Java, down from 31%
- 20% use R, up from 18%
While Linux remains the most popular OS, Mac and Windows usage is growing rapidly.
- 74% use Linux/Unix, down from 75% in 2015
- 32% use Windows, up from 23%
- 22% use Mac OSX, up from 14%

The report also includes statistics about the Spark community at large.

— Databricks reports growth in the contributor base from 600 in 2015 to 1,000 in 2016, a figure that does not seem to square with the statistics reported in OpenHub.

— Spark Meetup membership grew from 66,000 in 2015 to 225,000 in 2016.

— Spark Summit attendance grew from 3,912 to 5,100.

For a copy of the report and an infographic, go here.

Big Analytics Roundup (March 14, 2016)

HPE wins the internet this week by announcing the re-re-release of Haven, this time on Azure. The other big story this week: Flink announces Release 1.0.

Third Time’s a Charm

Hewlett Packard Enterprise (HPE) announces Haven on Demand on Microsoft Azure; PR firestorm ensues. Haven is a ~~loose bundle of software assets salvaged from the train wreck of Autonomy, Vertica, ArcSight and HP Operations Management~~ machine learning suite, originally branded as HAVEn and announced by HP in June, 2013. Since then, the software hasn’t exactly gone viral; Haven failed to make KDnuggets’ list of the top 50 machine learning APIs last December, a list that includes the likes of Ersatz, Hutoma and Skyttle.

One possible reason for the lack of virality: although several analysts described Haven as “open source”, HP did not release the Haven source code, and did not offer the software under an open source license.

Other than those two things, it’s open source.

In 2015, HP released Haven on Helion Public Cloud, HP’s failed cloud platform.

So this latest announcement is a re-re-release of the software. On paper, the library looks like it has some valuable capabilities in text, images video and audio analytics. The interface and documentation look a bit rough, but, after all, this is a ~~first~~ third release.

Jim’s Latest Musings

Angus Loten of the WSJ’s CIO Journal interviews SAS CEO Jim Goodnight, who increasingly sounds like your great-uncle at Thanksgiving dinner, the one who complains about “these kids today.” Goodnight compares cloud computing to mainframe time sharing. That’s ironic, because although SAS runs in AWS, it does not offer elastic pricing, the one thing that modern cloud computing shares with timesharing.

Goodnight also pooh-poohs IoT, noting that “we don’t have any major IoT customers, and I haven’t seen a good example of IoT yet.” SAS’ Product Manager for IoT could not be reached for comment.

Meanwhile, SAS held its annual analyst conference at a posh resort in Steamboat Springs, Colorado; in his report for Ventana Research, David Menninger gushes.

Herbalife Messes Up, Blames Data Scientists

Herbalife discloses errors reporting non-financial information, blames “database scripting errors.” The LA Times reports; Kaiser Fung comments.

Explainers

— Several items from the morning paper this week:

Adrian Colyer explains CryptoNets, a combination of Deep Learning and homohorphic encryption. By encrypting your data before you load it into the cloud, you make it useless to a hacker.
Adrian explains Neural Turing Machines.
Adrian explains Memory Networks.
Citing a paper published by Google last year, Adrian explains why using personal knowledge questions for account recovery is a really bad thing.

— Data Artisans’ Robert Metzger explains Apache Flink.

— In a video, Eric Kramer explains how to leverage patient data with Dataiku Data Science Studio.

Perspectives

— In InfoWorld, Serdar Yegulalp examines Flink 1.0 and swallows whole the argument that Flink’s “pure” streaming is inherently superior to Spark’s microbatching.

— On the MapR blog, Jim Scott offers a more balanced view of Flink, noting that streaming benchmarks are irrelevant unless you control for processing semantics and fault tolerance. Scott is excited about Flink ease of use and CEP API.

— John Leonard interviews Vincent de Lagabbe, CTO of bitcoin tracker Kaiko, who argues that Hadoop is unnecessary if you have less than a petabyte of data. Lagabbe prefers Datastax Enterprise.

— Also in InfoWorld, Martin Heller reviews Azure Machine Learning, finds it too hard for novices. I disagree. I used AML in a classroom lab, and students were up and running in minutes.

Open Source Announcements

— Flink announces Release 1.0. DataArtisans celebrates.

Teradata Watch

CEO Mike Koehler demonstrates confidence in TDC’s future by selling 11,331 shares.

Commercial Announcements

— Objectivity announces that Databricks has certified ThingSpan, a graph analytics platform, to work with Spark and HDFS.

— Databricks announces that adtech company Sellpoints has selected the Databricks platform to deliver a predictive analytics product.

Big Analytics Roundup (February 29, 2016)

Happy Leap Day. Tachyon’s rebranding as Alluxio, release of CaffeOnSpark and GA for Google Cloud Dataproc lead the hard news this week. The Alluxio announcement has inspired big thinkers to share big thoughts. And, we have a nice crop of explainers. Scroll down to the bottom for another SQL on Hadoop benchmark.

Explainers

— In SearchDataManagement, Jack Vaughn explains Spark 2.0.

— In Datanami, Alex Woodie explains Structured Streaming in Spark 2.0.

— MapR’s Jim Scott explains Spark accumulators. Jim also explains Spark Streaming.

— DataArtisans’ Fabian Hueske introduces Flink.

— In SlideShare, Julian Hyde explains streaming SQL.

— Wes McKinney explains why pandas users should be excited about Apache Arrow.

— On her blog, Paige Roberts explains Project Tungsten, complete with pictures.

— Someone from Dremio explains Drillix, which is what you get when you combine Apache Phoenix and Apache Drill. (h/t Hadoop Weekly).

Perspectives

— In TheNextPlatform, Timothy Prickett Morgan argues that Tachyon Caching (Alluxio) is bigger than Spark

— In SiliconAngle, Maria Deutscher opines that Alluxio (née Tachyon) could replace HDFS for Spark users.

— In The New Stack, Susan Hall speculates that Apache Arrow’s columnar data layer could accelerate Spark and Hadoop. She means Hadoop in a general way, e.g. the Hadoop ecosystem.

— On the Dataiku blog, “Caroline” interviews John Kelly, Managing Director of Berkeley Research Group and asks him questions about data science. Left unanswered: is it “Data-ikoo” or “Day-tie-koo?”

— Alpine Data Labs’ Steven Hillion ruminates on success. He’d be better off ruminating on “how to raise your next round of venture capital.”

— Max Slater-Robins opines that Microsoft is inventing the future, which is even better than winning the internet.

— In ZDNet, Andrew Brust wonders if Databricks is vying for a full analytics stack, citing the new Dashboard feature as cause for wonder. He’s just trolling.

— In Search Cloud Applications, Joel Shore opines that streaming analytics is replacing complex event processing, which makes sense. He further opines that Flink will displace Spark for streaming, which doesn’t make sense. Shore interviews IBM’s Nagui Halim about streaming here.

Open Source Announcements

— Alluxio (née Tachyon) announces Release 1.0.0. Alluxio is open source software distributed through Git under an Apache license, but is not an Apache project. Yet. Release 1.0 includes frameworks for MapReduce, Spark, Flink and Zeppelin. Daniel Gutierrez reports.

— Yahoo releases CaffeOnSpark, a distributed deep learning package. Caffe is one of the better-known deep learning packages, with a track record in image recognition. Software is available on Git. For more information, see the Wiki. Alex Handy reports; Charlie Osborne reports.

— RapidMiner China announces availability of an extension for deep learning engine DL4J. The extension is open source, and works with the open source version of RapidMiner. DL4J sponsor Skymind collaborated.

Commercial Announcements

–Tachyon Nexus, the commercial venture founded to support Tachyon, the memory-centric virtual distributed storage system, announces that it has rebranded as Alluxio.

— Google announces general availability for its Cloud Dataproc managed service for Spark and Hadoop.

Funding Announcements

Health analytics vendor Health Catalyst lands a $70M Series E round.

AtScale Benchmarks SQL-on-Hadoop Engines

On the AtScale blog, Trystan Leftwich summarizes results from a benchmark test of Hive on Tez (1.2/0.7), ~~Cloudera~~ Apache Impala (2.3) and Spark SQL (1.6). The AtScale team tested Impala and Spark with Parquet and Hive on Tez with ORC. For test cases, the team used TPC-H data arranged in a star schema, and ran 13 queries in each SQL engine multiple times, averaging the results.

While Hortonworks recommends ORC with Hive/Tez, there are published cases where users achieved good results with Hive/Tez on Parquet. Since the storage format has a big impact on SQL performance, I would have tested Hive/Tez on Parquet as well. AtScale did not respond to queries on this point.

Key findings:

All three engines performed about the same on single-table queries, and on queries joining three small tables.
Spark and Impala ran faster than Hive on queries joining three large tables.
Spark ran faster than Impala on queries joining four or more tables.

The team ran the same tests with AtScale’s commercial caching technology, with significant performance improvements for all three engines.

In concurrency testing, Impala performed much better than Hive or Spark.

Details of the test available in a white paper here (registration required).

Big Analytics Roundup (November 16, 2015)

Just three main stories this week: possible trouble for a pair of analytic startups; Google releases TensorFlow to open source; and H2O delivers new capabilities at its annual meeting.

In other news, the Spark team announces Release 1.5.2, a maintenance release; and the Mahout guy announces Release 0.11.1, with bug fixes and performance improvements. (h/t Hadoop Weekly)

Two items of note from the Databricks blog:

— Darin McBeath describes Elsevier’s Spark use case and introduces spark-xml-utils, a Spark package contributed by his team. The package enables the Spark user to filter documents based on an Path expression, return specific nodes for an Path/XQuery expression and transform documents using an XLST stylesheet.

— Rachit Agarwal and Anurag Khandelwal of Berkeley’s AMPLab introduce Succinct, a distributed datastore for queries on compressed data. They announce release of Succinct Spark, a Spark package that enables search, count, range and random access queries on compressed RDDs. The authors claim a 75X performance advantage over native Spark using Succinct as a document store,

Three interesting stories on streaming data:

In a podcast, Data Artisans CTO Stephan Ewen discusses Flink, Spark and the Kappa architecture.
Techalpine’s Kaushik Pal compares Spark and Flink for streaming data.
Will McGinnis helps you get started with Python and Flink.

(1) Analytic Startups in Trouble

In The Information, Steve Nellis and Peter Schulz explain why startups return to the funding well frequently — and why those that don’t may be in trouble. Venture funding isn’t a perfect indicator of success, but is often the only indicator available. On the list: Skytree Software and Alpine Data Labs.

(2) Google Releases TensorFlow for Machine Learning

On the Google Research blog, Google announces open source availability of TensorFlow. TensorFlow is Google’s second generation machine learning system; it supports Deep Learning as well as any computation that can be expressed as a flow graph. Read this white paper for details of the system. At present, there are Python and C++ APIs; Google notes that the C++ API may offer some performance advantages.

Video intro here.

In Wired, Cade Metz reports; Erik T. Mueller dismisses; and Metz returns to note that Deep Learning can leverage GPUs, and that AI’s future is in data, as if we didn’t know these things already.

On Slate, Will Oremus feels the buzz.

On his eponymous blog, Sachin Joglekar explains how to do k-means clustering with TensorFlow.

Separately, in VentureBeat, Jordan Novet rounds up open source frameworks for Deep Learning.

(3) H2O.ai Releases Steam

It’s not a metaphor. At its second annual H2O World event, H2O releases Steam, an open source data science hub that bundles model selection, model management and model scoring into a single container for elastic deployment.

On the H2O Blog, Yotam Levy wraps Day One, Day Two and Day Three of the H2O World event. Speaker videos are here, slides here. (Registration required.) Some notable presentations:

— H2O: Tomas Nykodym on GLM; Mark Landry on GBM and Random Forests; Arno Candel on Deep Learning; Erin LaDell on Ensemble Modeling.

— Michal Malohlava of H2O and Richard Garris of Databricks explain how to run H2O on Databricks Cloud. Separately, Michal demonstrates Sparkling Water, a Spark package that enables a Spark user to call H2O algorithms; Nidhi Mehta leads a hands-on with PySparkling Water; and Xavier Tordoir of Data Fellas exhibits Interactive Genomes Clustering with Sparkling Water on the Spark Notebook.

— Szilard Pafka of Epoch summarizes his work to date benchmarking R, Python, Vowpal Wabbit, H2O, xgboost and Spark MLLib. As reported previously, Pafka’s benchmarks show that H2O and xgboost are the best performers; they are faster and deliver more accurate models.

As reported in last week’s roundup, H2O.ai also announces a $20 million “B” round.

Spark Summit Europe Roundup

The 2015 Spark Summit Europe met in Amsterdam October 27-29. Here is a roundup of the presentations, organized by subject areas. I’ve omitted a few less interesting presentations, including some advertorials from sponsors.

State of Spark

— In his keynoter, Matei Zaharia recaps findings from Databricks’ Spark user survey, notes growth in summit attendance, meetup membership and contributor headcount. (Video here). Enhancements expected for Spark 1.6:

Dataset API
DataFrame integration for GraphX, Streaming
Project Tungsten: faster in-memory caching, SSD storage, improved code generation
Additional data sources for Streaming

— Databricks co-founder Reynold Xin recaps the last twelve months of Spark development. New user-facing developments in the past twelve months include:

DataFrames
Data source API
R binding and machine learning pipelines

Back-end developments include:

Project Tungsten
Sort-based shuffle
Netty-based network

Of these, Xin covers DataFrames and Project Tungsten in some detail. Looking ahead, Xin discusses the Dataset API, Streaming DataFrames and additional Project Tungsten work. Video here.

Getting Into Production

— Databricks engineer and Spark committer Aaron Davidson summarizes common issues in production and offers tips to avoid them. Key issues: moving beyond Python performance; using Spark with R; network and CPU-bound workloads. Video here.

— Tuplejump’s Evan Chan summarizes Spark deployment options and explains how to productionize Spark, with special attention to the Spark Job Server. Video here.

— Spark committer and Databricks engineer Andrew Or explains how to use the Spark UI to visualize and debug performance issues. Video here.

— Kostas Sakellis and Marcelo Vanzin of Cloudera provide a comprehensive overview of Spark security, covering encryption, authentication, delegation and authorization. They tout Sentry, Cloudera’s preferred security platform. Video here.

Spark for the Enterprise

— Revisting Matthew Glickman’s presentation at Spark Summit East earlier this year, Vinny Saulys reviews Spark’s impact at Goldman Sachs, noting the attractiveness of Spark’s APIs, in-memory processing and broad functionality. He recaps Spark’s viral adoption within GS, and its broad use within the company’s data science toolkit. His wish list for Spark: continued development of the DataFrame API; more built-in formulae; and a better IDE for Spark. Video here.

— Alan Saldich summarizes Cloudera’s two years of experience working with Spark: a host of engineering contributions and 200+ customers (including Equifax, Barclays and a slide full of others). Video here. Key insights:

Prediction is the most popular use case
Hive is most frequently co-installed, followed by HBase, Impala and Solr.
Customers want security and performance comparable to leading relational databases combined with simplicity.

Data Sources and File Systems

— Stephan Kessler of SAP and Santiago Mola of Stratio explain Spark integration with SAP HANA Vora through the Data Sources API. (Video unavailable).

— Tachyon Nexus’ Gene Pang offers an excellent overview of Tachyon’s memory-centric storage architecture and how to use Spark with Tachyon. Video here.

Spark SQL and DataFrames

— Michael Armbrust, lead developer for Spark SQL, explains DataFrames. Good intro for those unfamiliar with the feature. Video here.

— For those who think you can’t do fast SQL without a Teradata box, Gianmario Spacagna showcases the Insight Engine, an application built on Spark. More detail about the use case and solution here. The application, which requires many very complex queries, runs 500 times faster on Spark than on Hive, and likely would not run at all on Teradata. Video here.

— Informatica’s Kiran Lonikar summarizes a proposal to use GPUs to support columnar data frames. Video here.

— Ema Orhian of Atigeo describes jaws, a restful data warehousing framework built on Spark SQL with Mesos and Tachyon support. Video here.

Spark Streaming

— Helena Edelson, VP of Product Engineering at Tuplejump, offers a comprehensive overview of streaming analytics with Spark, Kafka, Cassandra and Akka. Video here.

— Francois Garillot of Typesafe and Gerard Maas of virdata explain and demo Spark Streaming. Video here.

— Iulian Dragos and Luc Bourlier explain how to leverage Mesos for Spark Streaming applications. Video here.

Data Science and Machine Learning

— Apache Zeppelin creator and NFLabs co-founder Moon Soo Lee reviews the Data Science lifecycle, then demonstrates how Zeppelin supports development and collaboration through all phases of a project. Video here.

— Alexander Ulanov, Senior Research Scientist at Hewlett-Packard Labs, describes his work with Deep Learning, building on MLLib’s multilayer perceptron capability. Video here.

— Databricks’ Hossein Falaki offers an introduction to R’s strengths and weaknesses, then dives into SparkR. He provides an overview of SparkR architecture and functionality, plus some pointers on mixing languages. The SparkR roadmap, he notes, includes expanded MLLib functionality; UDF support; and a complete DataFrame API. Finally, he demos SparkR and explains how to get started. Video here.

— MLlib committer Joseph Bradley explains how to combine the strengths R, scikit-learn and MLlib. Noting the strengths of R and scikit-learn libraries, he addresses the key question: how do you leverage software built to support single-machine workloads in a distributed computing environment? Bradley demonstrates how to do this with Spark, using sentiment analysis as an example. Video here.

— Natalino Busa of ING offers an introduction to real-time anomaly detection with Spark MLLib, Akka and Cassandra. He describes different methods for anomaly detection, including distance-based and density-based techniques. Video here.

— Bitly’s Sarah Guido explains topic modeling, using Spark MLLib’s Latent Dirchlet Allocation. Video here.

— Casey Stella describes using word2vec in MLLib to extract features from medical records for a Kaggle competition. Video here.

— Piotr Dendek and Mateusz Fedoryszak of the University of Warsaw explain Random Ferns, a bagged form of Naive Bayes, for which they have developed a Spark package. Video here.

GeoSpatial Analytics

— Ram Sriharsha touts Magellan, an open source geospatial library that uses Spark as an engine. Magellan, a Spark package, supports ESRI format files and GeoJSON; the developers aim to support the full suite of OpenGIS Simple Features for SQL. Video here.

Use Cases and Applications

— Ion Stoica summarizes Databricks’ experience working with hundreds of companies, distills to two generic Spark use cases: (1) the “Just-in-Time Data Warehouse”, bypassing IT bottlenecks inherent in conventional DW; (2) the unified compute engine, combining multiple frameworks in a single platform. Video here.

— Apache committer and SKT engineer Yousun Jeong delivers a presentation documenting SKT’s Big Data architecture and a use case real-time analytics. SKT needs to perform real-time analysis of the radio access network to improve utilization, as well as timely network quality assurance and fault analysis; the solution is a multi-layered appliance that combines Spark and other components with FPGA and Flash-based hardware acceleration. Video here.

— Yahoo’s Ayman Farahat describes a collaborative filtering application built on Spark that generates 26 trillion recommendations. Training time: 52 minutes; prediction time: 8 minutes. Video here.

— Sujit Pal explains how Elsevier uses Spark together with Solr, OpenNLP to annotate documents at scale. Elsevier has donated the application, called SoDA, back to open source. Video here.

— Parkinson’s Disease affects one out of every 100 people over 60, and there is no cure. Ido Karavany of Intel describes a project to use wearables to track the progression of the illness, using a complex stack including pebble, Android, IOS, play, Phoenix, HBase, Akka, Kafka, HDFS, MySQL and Spark, all running in AWS. With Spark, the team runs complex computations daily on large data sets, and implements a rules engine to identify changes in patient behavior. Video here.

— Paula Ta-Shma of IBM introduces a real-time routing use case from the Madrid bus system, then describes a solution that includes kafka, Secor, Swift, Parquet and elasticsearch for data collection; Spark SQL and MLLib for pattern learning; and a complex event processing engine for application in real time. Video here.

Benchmark: Spark Beats MapReduce

A group of scientists affiliated with IBM and several universities report on a detailed analysis of MapReduce and Spark performance across four different workloads. In this benchmark, Spark outperformed MapReduce on Word Count, k-Means and Page Rank, while MapReduce outperformed Spark on Sort.

On the ADT Dev Watch blog Dave Ramel summarizes the paper, arguing that it “brings into question..Databricks Daytona GraySort claim”. This point refers to Databricks’ record-setting entry in the 2014 Sort Benchmark run by Chris Nyberg, Mehul Shah and Naga Govindaraju.

However, Ramel appears to have overlooked section 3.3.1 of the paper, where the researchers explicitly address this question:

This difference is mainly because our cluster is connected using 1 Gbps Ethernet, as compared to a 10 Gbps Ethernet in, i.e., in our cluster configuration network can become a bottleneck for Sort in Spark.

In other words, had they deployed Spark on a cluster with high-speed network connections, it likely would run the Sort faster than MapReduce did.

I guess we’ll know when Nyberg et. al. release the 2015 GraySort results.

The IBM benchmark team found that k-means ran about 5X faster in Spark than in MapReduce. Ramel highlights the difference between this and the Spark team’s claim that machine learning algorithms run “up to” 100X faster.

The actual performance comparison shown on the Spark website compares logistic regression, which the IBM researchers did not test. One possible explanation — the Spark team may have tested against Mahout’s logistic regression algorithm, which runs on a single machine. It’s hard to say, since the Spark team provides no backup documentation for its performance claims. That needs to change.

Spark 1.5 Released

On September 9, the Spark team announced availability of Release 1.5. (Release notes here.) 230 developers contributed more than 1,400 commits, the largest release to date. Spark continues to expand its contributor base, the best measure of health for an open source project.

On the Databricks blog, Reynold Xin and Patrick Wendell summarize the key new bits: Some highlights:

Project Tungsten, a set of major changes to Spark’s internal architecture will be on by default. Spark 1.5 includes binary processing and a new code generation framework, with more than 100 built-in functions for common tasks.
Other performance enhancements include improved Parquet support (with predicate push-down and a faster metadata lookup path), and improved joins.
Usability enhancements include visualization of the SQL and DataFrame query plans in the web UI; the ability to connect to multiple versions of Hive metastores and the ability to read several Parquet variants.
Spark Streaming adds stability features, backpressure support, load balancing and several Python APIs.
The R interface is expanded to include Generalized Linear Models
New machine learning features include eight new transformers, three new estimators (naive Bayes, k-means and isotonic regression) plus three new algorithms (multilayer perceptron classifier, PrefixSpan for sequential pattern mining and FP-Growth for association rule learning)
Enhancements to existing algorithms include improvements to LDA, decision tree and ensemble features, an improved Pregel API for GraphX plus an ability to distribute matrix inversions for Gaussian Mixture Models (GMM).
Other new machine learning features include model summaries for linear and logistic regression, a splitting tool to define train and validation samples and a multiclass classification evaluator.

GraphX development has flatlined since the component graduated from Alpha in Spark 1.2.

Mesosphere, Typesafe, Tencent, Palantir, Cloudera, Hortonworks, Huawei, Shopify, Netflix, Intel, Yahoo, Kixer, UC Berkeley and Databricks all participated in release testing. Note that IBM, for all its marketing hoopla, contributes little or nothing to the project.

Spark Summit 2015: Preliminary Report

So I guess Spark really is enterprise ready. Nick Heudecker, call your office.

There are several key themes coming from the Summit:

Spark Continues to Mature

Spark and its contributors deserve a round of applause. Some key measures of growth since the 2014 Summit:

Contributor headcount increased from 255 to 730
Committed lines of code increased from 175K to 400K

There is increasing evidence of Spark’s scalability:

Largest cluster: 8,000 nodes
Largest job: 1 petabyte
Top streaming intake: 1TB/hour

Project Tungsten aims to make Spark faster and prepare for the next five years; the project has already accomplished significant performance improvements through better use of memory and CPU.

IBM and Spark

IBM drops the big one with its announcement. Key bits from the announcement:

IBM will build Spark into the core of its analytic and commerce products, including IBM Watson Health Cloud
IBM will open source its machine learning library (System ML) and work with Databricks to port it to Spark.
IBM will offer Spark as a Cloud service on Bluemix.
IBM will commit 3,500 developers to Spark-related projects.
IBM (and its partners) will train more than a million people on Spark

I will post separately on this next week

Spark is Enterprise-Ready

If IBM’s announcement is not sufficient to persuade skeptics, presentations from Adobe, Airbnb, Baidu, Capital One, CIA, NASA/JPL, NBC Universal, Netflix, Thompson Reuters, Toyota and many others demonstrate that Spark already supports enterprise-level workloads.

In one of the breakouts, Arsalan Tavakoli-Shiraji of Databricks presented results from his analysis of more than 150 production deployments of Spark. As expected, organizations use Spark for BI and advanced analytics; the big surprise is that 60% use non-HDFS data sources. These organizations use Spark for data consolidation on the fly, decoupling compute from storage, with unification taking place on the processing layer.

Databricks Cloud is GA

Enough said.

SparkR

Spark 1.4 includes R bindings, opening Spark to the large community of R users. Out of the gate, the R interface enables the R user to leverage Spark DataFrames; the Spark team plans to extend the capability to include machine learning APIs in Spark 1.5.

Spark’s Expanding Ecosystem

Every major Hadoop distributor showed up this year, but there were no major announcements from the distributors (other than IBM’s bombshell).

In other developments:

Amazon Web Services announced availability of a new Spark on EMR service
Intel announced a new Streaming SQL project for Spark
Lucidworks showcased its Fusion product, with Spark embedded
Alteryx announced its plans to integrate with Spark in its Release 10

One interesting footnote — while there were a number of presentations about Tachyon last year, there were none this year.

These are just the key themes. I’ll publish a more detailed story next week.