H2O (software)

H2O is open source software for big-data analysis . It is produced by the company H2O.ai (formerly 0xdata ), which launched in 2011 in Silicon Valley . H2O allows users to make thousands of potential models as part of discovering patterns in data.

H2O’s mathematical core is developed with the leadership of Arno Candel, part of Fortune’s 2014 “Big Data All Stars”. [6] The firm’s scientific advisors are experts on statistical learning theory and mathematical optimization .

The H2O software runs can be called from the R package , Python , and other environments. It is used for analyzing and analyzing data sets held in cloud computing systems and in the Apache Hadoop Distributed File System as well as in the Linux operating system , macOS , and Microsoft Windows . H2O The software is written in Java , Python , and R . Its graphical user interface is compatible with four browsers: Chrome , Safari , Firefox , andInternet Explorer .


The H2O project aims to develop an analytical interface for cloud computing, providing users with tools for data analysis. [1]


H2O’s chief executive, SriSatish Ambati, had helped to start Platfora , a big-data firm that develops software for the Apache Hadoop distributed file system. [7] Ambati Was frustrated with the performance of the R programming language on large data-sets and started the development of H2O software with encouragement from John Chambers , [2] Who created the S programming language at Bell Labs and who is a member of ‘ s core team (qui leads the development of R ). [2] [8] [9]

Ambati co-founded 0xdata with Cliff Click, who served the chief technical officer of H2O and helped create much of H2O’s product. Click to write the HotSpot Server Compiling and working with Azul Systems to build a big-data Java virtual machine (JVM). [10] Click left H2O in February 2016. [11] Leland Wilkinson , author of The Grammar of Graphics, serves as Chief Scientist and provides visualization leadership. [12]

Scientific advisory council

H2O’s Scientific Advisory Council lists three mathematical scientists, who are all professors at Stanford University: [13] Professor Stephen P. Boyd is an expert in convex minimization and applications in statistics and electrical engineering. [14] Robert Tibshirani , collaborator with Bradley Efron on bootstrapping , [15] is an expert on generalized additive models and statistical learning theory . [16] [17] Trevor Hastie , collaborator of John Chambers on S , [9] is an expert on generalized additive modelsand statistical learning theory. [16] [17]

H2O.ai: Silicon Valley start-up

Main article: H2O.ai

The software is open-source and distributed. The company receives fees for providing customer service and customized extensions. In November 2014, its customers included Cisco , eBay, Nielsen , and PayPal , according to VentureBeat . [2]

Mining of big data

Big data sets are too large to be Analyzed using traditional software like R . The H2O software provides data structures and methods suitable for big data. H2O allow users to analyze and visualize whole sets of data without using the Procrustean strategy of studying only a small subset with a conventional statistical package. [2] H2O’s statistical algorithms include K-means clustering , generalized linear models , distributed random forests , gradient boosting machines , naive bayes , major component analysis , and generalized low rank models . [18]

H2O is also able to run on Spark. [19]

Iterative methods for real-time problems

H2O uses iterative methods to provide quick answers. When a client can not wait for an optimal solution, the client can switch the computations and use an approximate solution. [1] In its approach to deep learning , [2] [18] [20] H2O divides all the data into subsets and then analyzes each subset simultaneously using the same method. These methods are combined to estimate parameters by using the Hogwild scheme, [21] a parallel stochastic gradient method. [22] These methods allow us to provide more information about the client than any other type of information.


Programming languages

The H2O software has an interface to the following programming languages: Java (6 or later), Python (2.7.x, 3.5.x), R (3.0.0 or later) and Scala (1.4-1.6). [2] [3]

Operating systems

The H2O software can be run on standard operating systems: Microsoft Windows ( 7 or later), Mac OS X ( 10.9 or later), and Linux ( Ubuntu 12.04 , RHEL / CentOS 6 or later), [3] It also runs on big-data systems, especially Apache Hadoop Distributed File System (HDFS), several popular versions: Cloudera (5.1 or later), MapR (3.0 or later), and Hortonworks (HDP 2.1 or later). It also operates on cloud computing environments, for example using Amazon EC2, Google Compute Engine , and Microsoft Azure . The H2O Sparkling Water Software is Databricks -certified on Apache Spark . [3]

Graphical user interface and browsers

Its graphical user interface is compatible with four browsers ( Chrome , Safari , Firefox , Internet Explorer ( IE10 ). [3]


  1. ^ Jump up to:c Harris (2012)
  2. ^ Jump up to:g Novet (2014)
  3. ^ Jump up to:f “Recommended systems for H2O” . 0xdata.com . H2O.ai. May 2015.
  4. Jump up^ Hardy (2014)
  5. Jump up^ https://github.com/h2oai/h2o-2/blob/master/LICENSE.txt
  6. Jump up^ Hackett (2014)
  7. Jump up^ Gage (2013)
  8. Jump up^ ACM honors Dr. John M. Chambers of Bell Labs with the 1998 ACM Software System Award for creating “S System” software, ACM press release, March 29, 1999. Accessed 8 December 2008.
  9. ^ Jump up to:b J. Chambers and T. Hastie, Statistical Models in S Wadsworth / Brooks Cole, 1991.
  10. Jump up^ Schuster, Werner (10 January 2014). “Cliff Click on in-memory processing, 0xdata H20, efficient low latency Java and GCs” . InfoQ . Retrieved 2 June 2015 .
  11. Jump up^ “Winds of Change” . Cliff Click. 2016.
  12. Jump up^ “H2O.ai” . www.h2o.ai . Retrieved 2017-01-28 .
  13. Jump up^ “About” . 0xdata. 2015.
  14. Jump up^ Boyd, Stephen P .; Vandenberghe, Lieven (2004). Convex optimization . Cambridge University Press. ISBN  978-0-521-83378-3 . Retrieved October 15, 2011 . (Free download of PDF of corrected 7th printing, 2009)
  15. Jump up^ Bradley Efron; Robert Tibshirani (1994). An Introduction to the Bootstrap . Chapman & Hall / CRC. ISBN  978-0-412-04231-7 .
  16. ^ Jump up to:b Hastie, TJ; Tibshirani, RJ (1990). Generalized additive models . Chapman & Hall / CRC. ISBN  978-0-412-34390-2 .
  17. ^ Jump up to:b Hastie, Trevor ; Tibshirani, Robert ; Friedman, Jerome H. (2011). The Elements of Statistical Learning (second ed.). Archived from the original on 10 November 2009 . Retrieved 15 June 2012 . ( Free download of 10th edition, June 2013 )
  18. ^ Jump up to:b Aiello, Spencer; Tom Kraljevic; Petr Maj (2015), with contributions from the 0xdata team, “h2o: R Interface for H2O” , The Comprehensive R Archive Network (CRAN) , Contributed Packages, The R Project for Statistical Computing (
  19. Jump up^ “FAQ – H2O documentation” . docs.h2o.ai . Retrieved 2017-01-28 .
  20. Jump up^ “Prediction of IncRNA using Deep Learning Approach”. Tripathi, Rashmi; Kumari, Vandana; Patel, Sunil; Singh, Yashbir; Varadwaj, Pritish. International Conference on Advances in Biotechnology (BioTech). Proceedings: 138-142. Singapore: Global Science and Technology Forum. (2015)
  21. Jump up^ Description of the iterative method for computing maximum-probability estimates for a generalized linear model.
  22. Jump up^ Benjamin Recht; Re, Christopher; Wright, Stephen & Feng Niu (2011). J. Shawe-Taylor; RS Zemel; PL Bartlett; F. Pereira & KQ Weinberger, eds. “Hogwild: A Lock-Free Approach to Parallelizing Stochastic Gradient Descent” (PDF) . Advances in Neural Information Processing Systems . Curran Associates, Inc. 24 : 693-701. Recht’s PDF


  • Gage, Deborah (April 15, 2013). “Platfora founder goes in search of big-data-answers” . Wall Street Journal . Retrieved 2 June 2015 .
  • Hackett, Robert (August 3, 2014), Nusca, Andrew; Hackett, Robert; Gupta, Shalene, eds., “Arno Candel, physicist and hacker, 0xdata” , Fortune , Meet Fortune’s 2014 Big Data All-Stars , retrieved 2 June 2015
  • Hardy, Quentin (3 May 2014). “Valuable humans in our digital future” . New York Times . Retrieved 1 June 2015 .
  • Harris, Derrick (August 14, 2012). “How 0xdata wants to help everyone become data scientists” . Gigaom Research . Retrieved 1 June 2015 .
  • Novet, Jordan (7 November 2014). “0xdata takes $ 8.9M and becomes H2O to match its open-source machine-learning project” . VentureBeat . Retrieved 1 June 2015 .