Big Data 7: yorkr waltzes with Apache NiFi

In this post, I construct an end-to-end Apache NiFi pipeline with my R package yorkr. This post is a mirror of my earlier post Big Data-5: kNiFing through cricket data with yorkpy based on my Python package yorkpy. The Apache NiFi Data Pipeilne flows all the way from the source, where the data is obtained, all the way to target analytics output. Apache NiFi was created to automate the flow of data between systems. NiFi dataflows enable the automated and managed flow of information between systems. This post automates the flow of data from Cricsheet, from where the zip file it is downloaded, unpacked, processed, transformed and finally T20 players are ranked.

This post uses the functions of my R package yorkr to rank IPL players. This is a example flow, of a typical Big Data pipeline where the data is ingested from many diverse source systems, transformed and then finally insights are generated. While I execute this NiFi example with my R package yorkr, in a typical Big Data pipeline where the data is huge, of the order of 100s of GB, we would be using the Hadoop ecosystem with Hive, HDFS Spark and so on. Since the data is taken from Cricsheet, which are few Megabytes, this approach would suffice. However if we hypothetically assume that there are several batches of cricket data that are being uploaded to the source, of different cricket matches happening all over the world, and the historical data exceeds several GBs, then we could use a similar Apache NiFi pattern to process the data and generate insights. If the data is was large and distributed across the Hadoop cluster , then we would need to use SparkR or SparklyR to process the data.

This is shown below pictorially

While this post displays the ranks of IPL batsmen, it is possible to create a cool dashboard using UI/UX technologies like AngularJS/ReactJS. Take a look at my post Big Data 6: The T20 Dance of Apache NiFi and yorkpy where I create a simple dashboard of multiple analytics

My R package yorkr can handle both men’s and women’s ODI, and all formats of T20 in Cricsheet namely Intl. T20 (men’s, women’s), IPL, BBL, Natwest T20, PSL, Women’s BBL etc. To know more details about yorkr see Revitalizing R package yorkr

The code can be forked from Github at yorkrWithApacheNiFi

You can take a look at the live demo of the NiFi pipeline at yorkr waltzes with Apache NiFi

Basic Flow

1. Overall flow

The overall NiFi flow contains 2 Process Groups a) DownloadAnd Unpack. b) Convert and Rank IPL batsmen. While it appears that the Process Groups are disconnected, they are not. The first process group downloads the T20 zip file, unpacks the. zip file and saves the YAML files in a specific folder. The second process group monitors this folder and starts processing as soon the YAML files are available. It processes the YAML converting it into dataframes before storing it as CSV file. The next processor then does the actual ranking of the batsmen before writing the output into IPLrank.txt

1.1 DownloadAndUnpack Process Group

This process group is shown below

1.1.1 GetT20Data

The GetT20Data Processor downloads the zip file given the URL

The ${T20data} variable points to the specific T20 format that needs to be downloaded. I have set this to https://cricsheet.org/downloads/ipl.zip. This could be set any other data set. In fact we could have parallel data flows for different T20/ Sports data sets and generate

1.1.2 SaveUnpackedData

This processor stores the YAML files in a predetermined folder, so that the data can be picked up by the 2nd Process Group for processing

1.2 ProcessAndRankT20Players Process Group

This is the second process group which converts the YAML files to pandas dataframes before storing them as. CSV files. The RankIPLPlayers will then read all the CSV files, stack them and then proceed to rank the IPL players. The Process Group is shown below

1.2.1 ListFile and FetchFile Processors

The left 2 Processors ListFile and FetchFile get all the YAML files from the folder and pass it to the next processor

1.2.2 convertYaml2DataFrame Processor

The convertYaml2DataFrame Processor uses the ExecuteStreamCommand which call Rscript. The Rscript invoked the yorkr function convertYaml2DataframeT20() as shown below

I also use a 16 concurrent tasks to convert 16 different flowfiles at once

library(yorkr)
args<-commandArgs(TRUE)
convertYaml2RDataframeT20(args[1], args[2], args[3])

1.2.3 MergeContent Processor

This processor’s only job is to trigger the rankIPLPlayers when all the FlowFiles have merged into 1 file.

1.2.4 RankT20Players

This processor is an ExecuteStreamCommand Processor that executes a Rscript which invokes a yorrkr function rankIPLT20Batsmen()

library(yorkr)
args<-commandArgs(TRUE)

rankIPLBatsmen(args[1],args[2],args[3])

1.2.5 OutputRankofT20Player Processor

This processor writes the generated rank to an output file.

1.3 Final Ranking of IPL T20 players

The Nodejs based web server picks up this file and displays on the web page the final ranks (the code is based on a good youtube for reading from file)

[1] "Chennai Super Kings"
[1] "Deccan Chargers"
[1] "Delhi Daredevils"
[1] "Kings XI Punjab"
[1] "Kochi Tuskers Kerala"
[1] "Kolkata Knight Riders"
[1] "Mumbai Indians"
[1] "Pune Warriors"
[1] "Rajasthan Royals"
[1] "Royal Challengers Bangalore"
[1] "Sunrisers Hyderabad"
[1] "Gujarat Lions"
[1] "Rising Pune Supergiants"
[1] "Chennai Super Kings-BattingDetails.RData"
[1] "Deccan Chargers-BattingDetails.RData"
[1] "Delhi Daredevils-BattingDetails.RData"
[1] "Kings XI Punjab-BattingDetails.RData"
[1] "Kochi Tuskers Kerala-BattingDetails.RData"
[1] "Kolkata Knight Riders-BattingDetails.RData"
[1] "Mumbai Indians-BattingDetails.RData"
[1] "Pune Warriors-BattingDetails.RData"
[1] "Rajasthan Royals-BattingDetails.RData"
[1] "Royal Challengers Bangalore-BattingDetails.RData"
[1] "Sunrisers Hyderabad-BattingDetails.RData"
[1] "Gujarat Lions-BattingDetails.RData"
[1] "Rising Pune Supergiants-BattingDetails.RData"
# A tibble: 429 x 4
   batsman     matches meanRuns meanSR
   <chr>         <int>    <dbl>  <dbl>
 1 DA Warner       130     37.9   128.
 2 LMP Simmons      29     37.2   106.
 3 CH Gayle        125     36.2   134.
 4 HM Amla          16     36.1   108.
 5 ML Hayden        30     35.9   129.
 6 SE Marsh         67     35.9   120.
 7 RR Pant          39     35.3   135.
 8 MEK Hussey       59     33.8   105.
 9 KL Rahul         59     33.5   128.
10 MN van Wyk        5     33.4   112.
# â€¦ with 419 more rows

Conclusion

This post demonstrated an end-to-end pipeline with Apache NiFi and R package yorkr. You can this pipeline and generated different analytics using the various functions of yorkr and display them on a dashboard.

Hope you enjoyed with post!

To see posts click Index of posts

Technology Trends – 2011 and beyond

There are lots of exciting things happening in the technological landscape. Innovation and development in every age is dependent on a set of key driving factors namely – the need for better, faster and cheaper, the need to handle disruptive technologies, the need to keep costs down and the need to absorb path breaking innovations. Given all these factors and the current trends in the industry the following technologies will enter mainstream in the years to come.

Long Term Evolution (LTE): LTE, also known as 4G technologies, has been born out of the disruptive entry of data hungry smart phones and tablet PCs. Besides, the need for better and faster applications has been the key driver of this technology. LTE is a data only technology that allows mobile users to access the internet on the move. LTE uses OFDM technology for sending and receiving data from user devices and also uses MIMO (multiple-in, multiple out). LTE is more economical, and spectrally efficient when compared to earlier 3.5G technologies like HSDPA, HSUPA and HSPA. LTE promises a better Quality of Experience (QoE) for end users.

IP Multimedia Systems IMS): IMS has been around for a while. However with the many advances in IP technology and the transport of media the time is now ripe for this technology to take wings and soar high. IMS uses the ubiquitous internet protocol for its core network both for media transport and for SIP signaling. Many innovative applications are possible with IMS including high definition video conferencing, multi-player interactive games, white boarding etc.

All senior management personnel of organizations are constantly faced with the need to keep costs down. The next two technologies hold a lot of promise in reducing costs for organizations and will surely play a key role in the years to come.

Cloud Computing: Cloud Computing obviates the need for upfront capital and infrastructure costs of organizations. Enterprises can deploy their applications on a public cloud which provides virtually infinite computing capacity in the hands of organizations. Organizations only pay as much as they use akin to utilities like electricity or water

Analytics: These days’ organizations are faced with a virtual deluge of data from their day to day operations. Whether the organizations belong to retail, health, finance, telecom, or transportation there is a lot of data that is generated. Data by itself is useless. This is where data analytics plays an important role. Predictive analytics help in classifying data, determining key trends and identifying correlations between data. This helps organizations in making strategic business decisions.

The following two technologies listed below are really path breaking and their applications are limitless.

Internet of Things: This technology envisages either passive or intelligent devices connected to the internet with a database at the back end for processing the data collected from these intelligent devices. This is also known as M2M (machine to machine) technology. The applications range from monitoring the structural integrity of bridges to implantable devices monitoring fatal heart diseases of patients.

Semantic Web (Web 3.0): This is the next stage in the evolution of the World Wide Web. The Web is now a vast repository of ideas, thoughts, blogs, observations etc. This technology envisages intelligent agents that can analyze the information in the web. These agents will determine the relations between information and make intelligent inferences. This technology will have to use artificial intelligence techniques, data mining and cloud computing to plumb the depths of the web

Conclusion: Creativity and innovation has been the hallmark of mankind from time immemorial. With the demand for smarter, cheaper and better the above technologies are bound to endure in the years to come.

Find me on Google+

The Future of Telecom

Published in Voice & Data – Bright Future

Introduction: The close of the 20^th century will long be remembered for one thing. The dotcom bust followed by the downward spiral of many major telecom and technology companies. For those who believe in the theory of the 12 year economic cycle this downturn is right about to end and we should see good times soon. Even otherwise there is good news for those in the telecom domain. We could shortly be witness to golden years ahead. There are many signs that seem to indicate that the telecom industry is on the verge of many major breakthroughs. Technologies like LTE, IMS, smartphones, cloud computing point to interesting times ahead. In fact telecom is at a inflexion point when the fortunes seem to be pointed northward. This article looks at some of the promising technologies which are going to bring back the sunshine to telecom.

3G Technologies –Better Quality of Experience (QoE): The auction of the 3G spectrum ended after 131 days of hectic bidding for this cutting edge telecom technology. 3G promises a whole new customer experience backed by extremely high data speeds. 3G promises download speed of up to 2 Mbps for stationary subscribers and 384 Kbps for moving subscribers. It is very clear that such high data speeds will inspire a host of new and exciting applications. Applications that span location based services (LBS), m-Commerce and NFC communications will be simply be irresistible to the users. Moreover the ability to watch video clips or live action on mobile TV or on laptops enabled with 3G dongles will have a lot of takers for 3G technology. App stores for 3G are bound to do a roaring business as 3G takes off in India.

Smartphones – The game changers: In the last decade or so in the telecom industry no other invention has had such a disruptive effect in the telecom domain as smartphones. Smartphones like the IPhone, Droid or Nexus One have changed the rules of the game. The impact of smartphone has been so huge that it actually spawned an entire industry of developers who developed applications for smartphones, content developers and app stores. The irresistible appeal of smartphones is the ease of use and the ability to browse the net as though they were using a normal data connection. Users can watch youtube clips, play games or chat on the Smartphone.

IP Multimedia Systems (IMS) – Digital Convergence: IP Multimedia System (IMS) , based on 3GPP’s Release 5 Specification in 2005, has been in the wings for quite some time. The IMS envisions an access agnostic telecommunication architecture that will use an all-IP Core for the transport of medium be it voice, data or video. IMS uses SIP protocol for signaling between network elements and SDP for exchanging media between applications. The IMS architecture promises a whole slew of exciting application ranging from high quality video conference, high speed data access, white boarding or real time interactive gaining. IMS represents a true convergence of the telecom wireless concepts with the data communication protocols. The types of services that are possible with IMS will be only limited by imagination. With the entry of smartphones and tablet PCs, IMS is a technology that is waiting to happen and will soon become prime time

Long Term Evolution (LTE) – Blazing Speeds: Already there are upward of 5 billion mobile devices and a report from Cisco states that the total data navigating the net will exceed ½ a zettabyte (10²¹) by the year 2013. The exponential growth of data and the need to provide even higher Quality of Experience (QoE) led to the development of the LTE. LTE is considered 4G technology. LTE promises speeds anywhere between to 56 Mbps to 100 Mbps to users enabling unheard of speeds and applications. What makes LTE so attractive is that it promises better spectral efficiency and lower cost per bit than 3G networks. The competing technology for LTE is WIMAX which is also considered as 4G. But LTE has a better evolution path from 3G networks as opposed to WiMAX, While LTE is a packet only network there are sound strategies for handling voice traffic with LTE. The standards body 3GPP offers two options for handling voice. The first is the Circuit switched (CS) fallback to 2G/3G network. In this scenario data access will be through the packet network of LTE while voice calls will use legacy 2G/3G voice networks. The other alternative is the switch voice traffic to the IMS network with its all-IP Core. This method is supported by the One Voice initiative of many major telecom companies and accepted by GSMA. This strategy for handling voice through an IMS network is known as VoLTE (Voice over LTE)

Internet of Things- Towards a connected World: “The Internet of Things” visualizes a highly interconnected world made of tiny passive or intelligent devices that connect to large databases and to the internet. This technology promises to transform the network from a dumb-bit pipe to a truly “computing” network. The Internet of Things or M2M (machine-to-machine) envisages an anytime, anywhere, anyone, anything network. The devices in this M2M network will be made up of passive elements, sensors and intelligent devices that communicate with the network. The devices will be capable of sensing, identifying and responding to changes in the immediate environment. Radio Frequency Identification (RFIDs) is one of the early and key enabler of this technology. The uses for this technology range from warning when the structural integrity of bridges is compromised to implantable devices in heart patients warning doctors of possible heart attacks. The impact of the Internet of Things will be far-reaching. There are numerous applications for this technology. In fact, ubiquitous computing or the Internet of Things allows us to distribute processing power and intelligence throughout the network into a kind of ambient intelligence spread across the network. This technology promises to blur the lines between science fiction and reality.

App Stores – The final verdict: The success of App Stores in the last couple of years has been nothing short of phenomenal. It is a complete ecosystem with App Store Developers, App Stores, and the Content Developers and Service Providers. Apps and App stores have changed the rules of the game so completely. No longer is a mobile phone’s snazzy looks enough for it to be a best seller. The mobile should be supported by cool downloadable apps for the user to use. App Stores and apps will play an increasingly important role with apps being developed for smartphones and tablet PCs. There are bound to be several interesting apps spanning technologies like Location Based Service (LBS), mobile Commerce, eTicketing, Near Field Communication

Cloud Computing – Utility computing: Cloud Computing has been around some but is slowly gaining more and more prominence. Cloud computing follows a utility model for computing where the cloud user only pays for the computing power and storage capacity used. Cloud computing not involve any upfront Capacity expenditure (Capex). Users of public clouds like EC2, App Engine or Azure can pay according to the usage of the resources provided by the cloud. Cloud technologies allow the CSPs to purchase processing power, platforms, and databases almost like a utility like electricity or water. The cloud exhibits an elastic behavior and expands to accommodate increasing demands and contracts when the demand drops. Cloud computing will be slowly be adopted by more and more organizations and enterprises in the years to come.

Analytics – Mining intelligence from data: Nowadays organizations all over are faced with a deluge of data. For raw data to be useful it has been analyzed, classified and important patterns determined from the data. This is where data mining and analytics come into play. Analytics uses statistical methods to classify data, determine correlations, identify patterns, and highlight and detect key trends among large data sets. Analytics enables industries to plumb the data sets through the process of selecting, exploring and modeling large amount of data to uncover previously unknown data patterns. The insights which analytics provides can be channelized to business advantage. Data mining and predictive analytics unlock the hidden secrets of data and help businesses make strategic decisions. Analytics is bound to become more common and will play a predominant role in all organizations in the years to come.

Internet TV – Hot off the net: If IMS represents the convergence of Telecom and the internet, Internet TV represents the marriage of TV and the internet. Internet TV is a technology whose time has come. Internet TV will bring a whole new user experience by allowing the viewer to be view rich content on his TV in an interactive manner. The technology titans like Apple, Microsoft and Google have their own version of this technology. Internet TV combines TV, the internet and apps for this new technology. Internet TV is bound to become popular with complementary technologies like IMS, LTE allowing for high speed data exchange and the popularity of websites like Youtube etc. Internet TV will receive a further boost from apps of smartphones and tablet PCs

IPv4 exhaustion – Damocles’ sword: While the future holds the promise of many new technologies it is also going throw a lot of attendant challenges. One serious problem that will need serious attention in the not too distant future is the IPv4 address space exhaustion. This problem may be even more serious than the Y2K problem. The issue is that IPv4 can address only 2 ³² or 4.3 billion devices. Already the pool has been exhausted because of new technologies like IMS which uses an all IP Core and the Internet of things with more devices, sensors connected to the internet – each identified by an IP address. The solution to this problem has been addressed long back and requires that the Internet adopt IPv6 addressing scheme. IPv6 uses 128-bit long address and allows 3.4 x 10³⁸ or 340 trillion, trillion, trillion unique addresses. However the conversion to IPv6 is not happening at the required pace and pretty soon will have to be adopted on war footing. It is clear that while the transition takes place, both IPv4 and IPv6 will co-exist so there will be an additional requirement of devices on the internet to be able to convert from one to another

Conclusion:

Technologies like IMS, LTE, and Internet TV have a lot of potential and hold a lot of promise. We as human beings have a constant need for better, faster and cheaper technologies. We can expect a lot of changes to happen in the next couple of years. We may once see rosy times ahead for telecom as a whole

<
Find me on Google+

The rise of analytics

Published in The Hindu – The rise of analytics

We are slowly, but surely, heading towards the age of “information overload”. The Sloan Digital Sky Survey started in the year 2000 returned around 620 terabytes of data in 11 months — more data than had ever been amassed in the entire history of astronomy.

The Large Hadron Collider (LHC) at CERN, Europe’s particle physics laboratory, in Geneva will during its search for the origins of the universe and the elusive Higgs particle, early next year, spew out terabytes of data in its wake. Now there are upward of five billion devices connected to the Internet and the numbers are showing no signs of slowing down.

A recent report from Cisco, the data networking giant, states that the total data navigating the Net will cross 1/2 a zettabyte (10 {+2} {+1}) by the year 2013.

Such astronomical volumes of data are also handled daily by retail giants including Walmart and Target and telcos such as AT&T and Airtel. Also, advances in the Human Genome Project and technologies like the “Internet of Things” are bound to throw up large quantities of data.

The issue of storing data is now slowly becoming non-existent with the plummeting prices of semi-conductor memory and processors coupled with a doubling of their capacity every 18 months with the inevitability predicted by Moore’s law.

Plumbing the depths

Raw data is by itself quite useless. Data has to be classified, winnowed and analysed into useful information before if it can be utilised. This is where analytics and data mining come into play. Analytics, once the exclusive preserve of research labs and academia, has now entered the mainstream. Data mining and analytics are now used across a broad swath of industries — retail, insurance, manufacturing, healthcare and telecommunication. Analytics enables the extraction of intelligence, identification of trends and the ability to highlight the non-obvious from raw, amorphous data. Using the intelligence that is gleaned from predictive analytics, businesses can make strategic game-changing decisions.

Analytics uses statistical methods to classify data, determine correlations, identify patterns, and highlight and detect deviations among large data sets. Analytics includes in its realms complex software algorithms such as decision trees and neural nets to make predictions from existing data sets. For e.g. a retail store would be interested in knowing the buying patterns of its consumers. If the store could determine that product Y is almost always purchased when product X is purchased then the store could come up with clever schemes like an additional discount on product Z when both products X & Y are purchased. Similarly, telcos could use analytics to identify predominant trends that promote customer loyalty.

Studying behaviour

Telcos could come with voice and data plans that attract customers based on consumer behaviour, after analysing data from its point of sale and retail stores. They could use analytics to determine causes for customer churn and come with strategies to prevent it.

Analytics has also been used in the health industry in predicting and preventing fatal infections in infants based on patterns in real-time data like blood pressure, heart rate and respiration.

Analytics requires at its disposal large processing power. Advances in this field have been largely fuelled by similar advances in a companion technology, namely cloud computing. The latter allows computing power to be purchased on demand almost like a utility and has been a key enabler for analytics.

Data mining and analytics allows industries to plumb the data sets that are held in the organisations through the process of selecting, exploring and modelling large amount of data to uncover previously unknown data patterns which can be channelised to business advantage.

Analytics help in unlocking the secrets hidden in data and provide real insights to businesses; and enable businesses and industries to make intelligent and informed choices.

In this age of information deluge, data mining and analytics are bound to play an increasingly important role and will become indispensable to the future of businesses.

Find me on Google+

Cloud, analytics key tools for today’s telcos

Published in Telecom Asia Aug 20, 2010 – http://bit.ly/dxKbsR

Operators facing dwindling revenue from wireline subscribers, fierce tariff wars and exploding mobile data traffic are continually being pressured to do more for less. Spending on infrastructure is increasing as they look to provide better service within slender budgets.

In these tough times telcos have to devise new and innovative strategies and make judicious technology choices. Two promising technologies, cloud computing and analytics, are shaping up as among the best choices to make.

Cloud architecture does away with the worry of planning the computing resources needed, the real estate, the costs of the acquiring them and thoughts of its obsolescence. It allows the CSPs to purchase processing power, platforms and databases almost as a utility like electricity or water.

Cloud consumers only pay for what they use. The magic of this promising technology is the elasticity that the cloud provides – it expands to accommodate increasing demands and contracts when the demand drops.

The cloud architectures of Amazon, Google and Microsoft – currently the three biggest cloud providers – vary widely in their capabilities and features. These strengths and weaknesses should be taken into account while planning a cloud system. Each is best suited for only a certain class of applications unique to each individual cloud provider.

On one end of the spectrum Amazon’s EC2 (Elastic Compute Cloud) provides a virtual machine and a wealth of associated tools for storage and notifications. But the trade-off for increased flexibility is that users must take responsibility for designing resiliency into their systems.

On the other end is Google’s App Engine, a highly scalable cloud architecture that handles failures but is a lot more restrictive. Microsoft’s Azure is based on the .NET architecture and in terms of flexibility and features lies between these two.

When implementing such architecture, an organization should take a long hard look its computing software inventory to decide which applications are worthy of migrating to the cloud. The best candidates are processing intensive in-house applications that deliver standardized functionality and interface, and whose software architecture is made up of loosely coupled communicating systems.

Applications that deal with sensitive data should be retained within the organization’s internal computing infrastructure, because security is currently the most glaring issue with the cloud. Cloud providers do provide various levels of security to users, but this is an area in keen need of standardization.

But if the CSP decides to build components of an OSS system – rather than buying a pre-packaged system – it makes good business sense to develop for the cloud.

A cloud-based application must have a few essential properties. First, it is preferable if the application was designed on SOA principles. Second, it should be loosely coupled. And lastly, it needs to be an application that can be scaled rapidly up or down based on the varying demands.

The other question is which legacy systems can be migrated. If the OSS/BSS systems are based on commercial off-the-shelf systems these can be excluded, but an offline bill processing system, for example, is typically a good candidate for migration.

Mining wisdom from data

The cloud can serve as the perfect companion for another increasingly vital operational practice – data analytics. The cloud is capable of modeling large amounts of data, and running models to process and analyze this data. It is possible to run thousands of simultaneous instances on the cloud and mine for business intelligence in the oceans of telecom data operators generate.

Today’s CSP maintains software systems generating all kinds of customer data, covering areas ranging from billing and order management to POS, VAS and provisioning. But perhaps the largest and richest vein of subscriber information is the call detail records database.

All this data is worthless, though, if it cannot be mined and analyzed. Formal data mining and data analytics tools can be used to identify patterns and trends that will allow operators to make strategic, knowledge-driven decisions.

Analytics involves many complex areas like predictive analytics, neural nets, decision trees and classification. Some of the approaches used in data analytics include prediction, deviation detection, degree of influence and classification.

With the intelligence that comes through analytics it is possible to determine customer buying patterns, identify causes for churn and develop strategies to promote loyalty. Call patterns based on demography or time of day will enable the CSPs to create innovative tariff schemes.

Determining the relations and buying patterns of users will provide opportunities for up-selling and cross-selling. The ability to identify marked deviation in customer behavior patterns help the CSP in deciding ahead of time whether this trend is a warning bell or an opportunity waiting to be tapped.

Tinniam V Ganesh

Find me on Google+