Slides for presentation at Data Science DC Meetup (a highly recommended group!) 2012-10-17 http://www.meetup.com/Data-Science-DC/events/83813992/

1. Building Enterprise Apps
for Big Data with Cascading
Paco Nathan
Document
Collection
Scrub
Tokenize
token
Concurrent, Inc.
M
HashJoin Regex
Left token
GroupBy R
Stop Word token
List
RHS
pnathan@concurrentinc.com Count
@pacoid
Word
Count
Copyright @2012, Concurrent, Inc.

2. Enterprise Apps
for Big Data
with Cascading
1. backstory: how we got here
2. build: Data Science teams
3. pattern: common use cases
4. intro: Cascading API
5. tutorial: for the impatient
6. code: sample apps

3. Intro to Cascading
Document
Collection
Scrub
Tokenize
token
M
HashJoin Regex
Left token
GroupBy R
Stop Word token
List
RHS
Count
Word
Count
1. backstory:
how we got here

4. inflection point
huge Internet successes after 1997 holiday season… 1997
AMZN, EBAY, Inktomi (YHOO Search), then GOOG
1998
consider this metric:
annual revenue per customer / amount of data stored
which dropped 100x within a few years after 1997 2004
storage and processing costs plummeted, now we must
work much smarter to extract ROI from Big Data…
our methods must adapt
“conventional wisdom” of RDBMS and BI tools became
less viable; however, business cadre was still focused on
pivot tables and pie charts… which tends toward inertia!
MapReduce and the Hadoop open source stack grew
directly out of that contention… however, that effort ＋
only solves parts of the puzzle

5. inflection point: consequences
Geoffrey Moore (Mohr Davidow Ventures, author of Crossing The Chasm)
Hadoop Summit, 2012:
“All of Fortune 500 is now on notice over the next 10-year period.”
Amazon and Google as exemplars of massive disruption in retail,
advertising, etc.
data as the major force displacing Global 1000 over the next decade,
mostly through apps — verticals, leveraging domain expertise
Michael Stonebraker (INGRES, PostgreSQL,Vertica,VoltDB, etc.)
XLDB, 2012:
“Complex analytics workloads are now displacing SQL as the basis
for Enterprise apps.”

6. primary sources
Amazon
“Early Amazon: Splitting the website” – Greg Linden
glinden.blogspot.com/2006/02/early-amazon-splitting-website.html
eBay
“The eBay Architecture” – Randy Shoup, Dan Pritchett
addsimplicity.com/adding_simplicity_an_engi/2006/11/you_scaled_your.html
addsimplicity.com.nyud.net:8080/downloads/eBaySDForum2006-11-29.pdf
Inktomi (YHOO Search)
“Inktomi’s Wild Ride” – Erik Brewer (0:05:31 ff)
youtube.com/watch?v=E91oEn1bnXM
Google
“The Birth of Google” – John Battelle
wired.com/wired/archive/13.08/battelle.html
“Underneath the Covers at Google” – Jeff Dean (0:06:54 ff)
youtube.com/watch?v=qsan-GQaeyk
perspectives.mvdirona.com/2008/06/11/JeffDeanOnGoogleInfrastructure.aspx

7. the world before…
BI, SQL, and highly
optimized code

8. data innovation: circa 1996
Stakeholder Customers
Excel pivot tables
PowerPoint slide decks strategy
BI
Product
Analysts
requirements
SQL Query optimized
Engineering code Web App
result sets
transactions
RDBMS

9. the world after…
machine learning,
leveraging log files

10. data innovation: circa 2001
Stakeholder Product Customers
dashboards UX
Engineering
models servlets
recommenders
Algorithmic + Web Apps
Modeling classifiers
Middleware
aggregation
event
SQL Query history
result sets customer
transactions
Logs
DW ETL RDBMS

11. the world ahead…
what our customers
are doing now

12. data innovation: circa 2013
Customers
Data Apps
business
Domain process Workflow Prod
Expert
dashboard Web Apps,
metrics
History services Mobile,
data etc. s/w
science dev
Data
Planner
Scientist
social
discovery optimized interactions
+ capacity transactions, Eng
endpoints
modeling content
App Dev
Data Access Patterns
Hadoop, Log In-Memory
etc. Events Data Grid
Ops DW Ops
batch "real time"
Cluster Scheduler
introduced existing
capability SDLC
RDBMS
RDBMS

13. a key difference…

14. statistical thinking
Process Variation Data Tools
employing a mode of thought which includes both logical and analytical reasoning:
evaluating the whole of a problem, as well as its component parts; attempting
to assess the effects of changing one or more variables
this approach attempts to understand not just problems and solutions,
but also the processes involved and their variances
particularly valuable in Big Data work when combined with hands-on experience in
physics – roughly 50% of my peers come from physics or physical engineering…
programmers typically don’t think this way…
however, both systems engineers and data scientists must!

15. reference
by Leo Breiman
Statistical Modeling:
The Two Cultures
Statistical Science, 2001
bit.ly/eUTh9L

16. Intro to Cascading
Document
Collection
Scrub
Tokenize
token
M
HashJoin Regex
Left token
GroupBy R
Stop Word token
List
RHS
Count
Word
Count
2. build:
Data Science teams

17. core values
Data Science teams develop actionable insights,
building confidence for decisions
that work may influence a few decisions worth
billions (e.g., M&A) or billions of small decisions
(e.g., AdWords)
probably somewhere in-between…
Wikipedia
solving for pattern, at scale.
an interdisciplinary pursuit which
requires teams, not sole players

18. most valuable skills
approximately 80% of the costs for data-related projects
get spent on data preparation – mostly on cleaning up
data quality issues: ETL, log file analysis, etc.
unfortunately, data-related budgets for many companies tend
to go into frameworks which can only be used after clean up
most valuable skills:
‣ learn to use programmable tools that prepare data
‣ learn to generate compelling data visualizations
‣ learn to estimate the confidence for reported results
‣ learn to automate work, making analysis repeatable
D3
the rest of the skills – modeling,
algorithms, etc. – those are secondary

19. social caveats
“This data cannot be correct!” may be an early warning
about an organization itself
much depends on how the people whom you work alongside
tend to arrive at decisions:
‣ probably good: Induction, Abduction, Circumscription
‣ probably poor: Deduction, Speculation, Justification
in general, one good data visualization
puts many ongoing verbal arguments to rest
however, let domain experts handle
“data storytelling”, not data scientists
xkcd

20. the science in data science?
edoMpUsserD:IUN
tcudorP ylppA lenaP yrotnevnI tneilC
tcudorP evomeR lenaP yrotnevnI tneilC
in a nutshell, what we do…
edoMmooRyM:IUN
edoMmooRcilbuP:IUN
ydduB ddA
nigoL etisbeW
vd
edoMsdneirF:IUN
edoMtahC:IUN
egasseM a evaeL
G1 :gniniamer ecaps sserddA
dekcilCeliforPyM:IUN
edoMstiderCyuB:IUN
tohspanS a ekaT
egapemoH nwO tisiV
elbbuB a epyT
taeS egnahC
wodniW D3 nepO
dneirF ddA
revO tcudorP pilF lenaP yrotnevnI tneilC
lenaP tidE
‣ estimate probability
woN tahC
teP yalP
teP deeF
2 petS egaP traC esahcruP edaM remotsuC
M215 :gniniamer ecaps sserddA
gnihtolC no tuP
bew :metI na yuB
edoMeivoM:IUN
ytinummoc ,tneilc :detratS weiV eivoM
teP weN etaerC
detrats etius tset :tseTytivitcennoC
emag pazyeh dehcnuaL
eciov mooRcilbuP tahC
egasseM yadhtriB
edoMlairotuT:IUN
ybbol semag dehcnuaL
noitartsigeR euqinU
‣ calculate analytic variance
edoMpUsserD:IUN
tcudorP ylppA lenaP yrotnevnI tneilC
tcudorP evomeR lenaP yrotnevnI tneilC
edoMmooRyM:IUN
edoMmooRcilbuP:IUN
ydduB ddA
nigoL etisbeW
vd
edoMsdneirF:IUN
edoMtahC:IUN
egasseM a evaeL
G1 :gniniamer ecaps sserddA
dekcilCeliforPyM:IUN
edoMstiderCyuB:IUN
tohspanS a ekaT
egapemoH nwO tisiV
elbbuB a epyT
taeS egnahC
dneirF ddA
revO tcudorP pilF lenaP yrotnevnI tneilC
lenaP tidE
woN tahC
teP yalP
teP deeF
2 petS egaP traC esahcruP edaM remotsuC
M215 :gniniamer ecaps sserddA
gnihtolC no tuP
bew :metI na yuB
edoMeivoM:IUN
ytinummoc ,tneilc :detratS weiV eivoM
teP weN etaerC
detrats etius tset :tseTytivitcennoC
emag pazyeh dehcnuaL
eciov mooRcilbuP tahC
egasseM yadhtriB
edoMlairotuT:IUN
ybbol semag dehcnuaL
noitartsigeR euqinU
wodniW D3 nepO
‣ manipulate order complexity
‣ make use of learning theory
+ collab with DevOps, Stakeholders
+ reduce our work to cron entries

21. synthesis of the above
MapReduce is Good Enough?
Jimmy Lin, U Maryland + Twitter
arxiv.org/pdf/1209.2191v1.pdf
A Few Useful Things to Know about Machine Learning
Pedro Domingos, U Washington
homes.cs.washington.edu/~pedrod/papers/cacm12.pdf

22. team process = needs
help people ask the
discovery right questions
allow automation to place
modeling informed bets
deliver products at
integration scale to customers
build smarts into
apps product features Gephi
keep infrastructure
systems running, cost-effective

23. team composition = roles
Domain
Expert
business process,
stakeholder
data
science
Data data prep, discovery,
Scientist modeling, etc. Document
Collection
Scrub
Tokenize
token
M
HashJoin Regex
Left token
GroupBy R
Stop Word token
List
RHS
App Dev
software engineering, Count
automation Word
Count
Ops systems engineering, access
introduced
capability

24. matrix = needs × roles
nn
o
overy
very elliing
e ng ratiio
rat o apps
apps tem
tem
ss
diisc
d sc mod
mod nteg
ii nteg sys
sys
stakeholder
scientist
developer
ops

25. matrix: example team
nn
o
overy
very elliing
e ng ratiio
rat o apps
apps tem
tem
ss
diisc
d sc mod
mod nteg
ii nteg sys
sys
stakeholder
scientist
developer
ops
summary: this team seems heavy on systems, may need more overlap
between modeling and integration, particularly among team leads

26. Q:
Can I simply hire one
rockstar data scientist
to cover all this work?

27. A: No, interdisciplinary
work requires teams.
A: Hire leads who speak
the lingo of each domain.
A: Hire people who cover
2+ roles, when possible.

28. reference
by DJ Patil
Data Jujitsu
O’Reilly, 2012
amazon.com/dp/B008HMN5BE
Building Data Science Teams
O’Reilly, 2011
amazon.com/dp/B005O4U3ZE

29. Intro to Cascading
Document
Collection
Scrub
Tokenize
token
M
HashJoin Regex
Left token
GroupBy R
Stop Word token
List
RHS
Count
Word
Count
3. pattern:
common use cases

30. CAP theorem
purpose: theoretical limits for data access patterns
essence:
‣ consistency
‣ availability
‣ partition tolerance
best case scenario: you may pick two … or spend billions
struggling to obtain all three at scale (GOOG)
translated: cost of doing business
www.cs.berkeley.edu/~brewer/cs262b-2004/PODC-keynote.pdf
julianbrowne.com/article/viewer/brewers-cap-theorem

31. data access patterns
design patterns: originated in consensus negotiation
for architecture, later used in software engineering
consider the corollaries in large-scale data work…
essence: select data frameworks based on
your data access patterns
in other words, decouple use cases based on needs
– avoid the “one size fits all” (OSFA) anti-pattern
let’s review some examples…

32. access → frameworks → forfeits
financial transactions general ledger in RDBMS CAx
ad-hoc queries RDS (hosted MySQL) CAx
reporting, dashboards like Pentaho CAx
log rotation/persistence like Riak xxP
search indexes like Lucene/Solr xAP
static content, archives S3 (durable storage) xAP
customer facts like Redis, Membase xAP
distributed counters, locks, sets like Redis x A P*
data objects CRUD key/value – like, NoSQL on MySQL CxP
authoritative metadata like Zookeeper CxP
data prep, modeling at scale like Hadoop/Cascading + R CxP
graph analysis like Hadoop + Redis + Gephi CxP
data marts like Hadoop/HBase CxP

33. access → frameworks → forfeits
financial transactions general ledger in RDBMS CAx
ad-hoc queries RDS (hosted MySQL) CAx
reporting, dashboards like Pentaho CAx
log rotation/persistence like Riak xxP
search indexes like Lucene/Solr xAP
static content, archives S3 (durable storage) xAP
customer facts like Redis, Membase xAP
distributed counters, locks, sets like Redis x A P*
data objects CRUD key/value – like, NoSQL on MySQL CxP
authoritative metadata like Zookeeper CxP
data prep, modeling at scale like Hadoop/Cascading + R CxP
graph analysis like Hadoop + Redis + Gephi CxP
data marts like Hadoop/HBase CxP

34. and, since
“One Size Fits All”…
doesn’t

35. a selection of great tools…
reporting:
visualization:
Graphite, PowerPivot,
ggplot2, D3, Gephi
analytics/modeling: Pentaho, Jaspersoft, SAS
R, Weka, Matlab, PMML, GLPK
text:
LDA, WordNet, OpenNLP, Mallet, Bixo, NLTK
apps:
Cascading, Scalding, Cascalog, R markdown, SWF
scale-out:
Scalr, RightScale, CycleComputing, vFabric, Beanstalk
graph: column:
Gremlin, Vertica,
GraphLab, HBase, key/val: index: relational:
Neo4J Drill, Redis, Lucene/Solr, usual suspects
Dynamo Membase, ElasticSearch
MySQL
imdg:
Spark, Storm, hadoop:
EMR, HW, MapR, machine data:
Gigaspaces
EMC, Azure, Compute Splunk, collectd, durable storage:
Nagios S3, ASV, GCS,
Riak, Couch

36. common use cases
app patterns

37. use case: marketing funnel
• must optimize a very large ad spend
• different vendors report different metrics
Wikipedia
• seasonal variation distorts performance
• some campaigns are much smaller than others
• hard to predict ROI for incremental spend
approach:
• log aggregation, followed with cohort analysis
• bayesian point estimates compare different-sized ad tests
• customer lifetime value quantifies ROI of new leads
• time series analysis normalizes for seasonal variation
• geolocation adjusts for regional cost/benefit
• linear programming models estimate elasticity of demand

38. use case: ecommerce fraud
• sparse data means lots of missing values
stat.berkeley.edu
• “needle in a haystack” lack of training cases
• answers are available in large-scale batch, results
are needed in real-time event processing
• not just one pattern to detect – many, ever-changing
approach:
• random forest (RF) classifiers predict likely fraud
• subsampled data to re-balance training sets
• impute missing values based on density functions
• train on massive log files, run on in-memory grid
• adjust metrics to minimize customer support costs
• detect novelty – report anomalies via notifications

39. use case: customer segmentation
• many millions of customers, hard to determine
which features resonate
Mathworks
• multi-modal distributions get obscured by the
practice of calculating an “average”
• not much is known about individual customers
approach:
• connected components for sessionization, determining
uniques from logs
• estimates for age, gender, income, geo, etc.
• clustering algorithms to group into market segments
• social graph infers “unknown” relationships
• covariance/heat maps visualizes segments vs. feature sets

40. use case: monetizing content
• need to suggest relevant content which would
Digital Humanities
otherwise get buried in the back catalog
• big disconnect between inventory and limited
performance ad market
• enormous amounts of text, hard to categorize
approach:
• text analytics glean key phrases from documents
• hierarchical clustering of char frequencies detects lang
• latent dirichlet allocation (LDA) reduces dimension to
topic models
• recommenders suggest similar topics to customers
• collaborative filters connect known users with less known

41. Intro to Cascading
Document
Collection
Scrub
Tokenize
token
M
HashJoin Regex
Left token
GroupBy R
Stop Word token
List
RHS
Count
Word
Count
4. intro:
Cascading API

42. Cascading API: purpose
‣ simplify data processing development and deployment
‣ improve application developer productivity
‣ enable data processing application manageability

43. Cascading API: a few facts
Java open source project (ASL 2) using Git, Gradle, Maven, JUnit, etc.
in production (~5 yrs) at hundreds of enterprise Hadoop deployments:
Finance, Health Care, Transportation, other verticals
studies published about large use cases: Twitter, Etsy, Airbnb, Square,
Climate Corporation, FlightCaster, Williams-Sonoma
partnerships and distribution with SpringSource, Amazon AWS,
Microsoft Azure, Hortonworks, MapR, EMC
several open source projects built atop, managed by Twitter, Etsy, etc.,
which provide substantial Machine Learning libraries
DSLs available in Scala, Clojure, Python (Jython), Ruby (JRuby), Groovy
data “taps” integrate popular data frameworks via JDBC, Memcached, HBase,
plus serialization in Apache Thrift, Avro, Kyro, etc.
entire app compiles into a single JAR: fully connected for compiler optimization,
exception handling, debugging, config, scheduling, etc.

44. Cascading API: a few quotes
“Cascading gives Java developers the ability to build Big Data applications
on Hadoop using their existing skillset … Management can really go out
and build a team around folks that are already very experienced with Java.
Switching over to this is really a very short exercise.”
CIO, Thor Olavsrud, 2012-06-06
cio.com/article/707782/Ease_Big_Data_Hiring_Pain_With_Cascading
“Masks the complexity of MapReduce, simplifies the programming, and
speeds you on your journey toward actionable analytics … A vast
improvement over native MapReduce functions or Pig UDFs.”
2012 BOSSIE Awards, James Borck, 2012-09-18
infoworld.com/slideshow/65089
“Company’s promise to application developers is an opportunity to build
and test applications on their desktops in the language of choice with
familiar constructs and reusable components”
Dr. Dobb’s, Adrian Bridgwater, 2012-06-08
drdobbs.com/jvm/where-does-big-data-go-to-get-data-inten/240001759

45. data+code “political spectrum”
“Notes from the Mystery Machine Bus”
by Steve Yegge, Google
goo.gl/SeRZa
“conservative” “liberal”
(mostly) Enterprise (mostly) Start-Up
risk management customer experiments
assurance flexibility
well-defined schema schema follows code
explicit configuration convention
type-checking compiler interpreted scripts
wants no surprises wants no impediments
Java, Scala, Clojure, etc. PHP, Ruby, Python, etc.
Cascading, Scalding, Cascalog, etc. Hive, Pig, Hadoop Streaming, etc.

46. Cascading API: adoption
As Enterprise apps move into
Hadoop and related BigData
frameworks, risk profiles shift
toward more conservative
programming practices
Cascading provides a popular
API for defining and managing
Enterprise data workflows

47. enterprise data workflows
Tuples, Pipelines, Endpoints, Operations, Joins, Assertions, Traps, etc.
…in other words, “plumbing”
Document
Collection
Scrub
Tokenize
token
M
HashJoin Regex
Left token
GroupBy R
Stop Word token
List
RHS
Count
Word
Count

48. data workflows: team
‣ Business Stakeholder POV:
business process management for workflow orchestration (think BPM/BPEL)
‣ Systems Integrator POV:
system integration of heterogenous data sources and compute platforms
‣ Data Scientist POV:
a directed, acyclic graph (DAG) on which we can apply Amdahl's Law, etc.
‣ Data Architect POV:
a physical plan for large-scale data flow management
‣ Software Architect POV:
a pattern language, similar to plumbing or circuit design
Document
Collection
‣ App Developer POV: M
Tokenize
Scrub
token
API bindings for Java, Scala, Clojure, Jython, JRuby, etc. Stop Word
List
HashJoin
Left
RHS
Regex
token
GroupBy
token
R
Count
‣ Systems Engineer POV: Word
Count
a JAR file, has passed CI, available in a Maven repo

49. data workflows: layers
business domain expertise, business trade-offs,
process operating parameters, market position, etc.
API Java, Scala, Clojure, Jython, JRuby, Groovy, etc.
language
…envision whatever runs in a JVM
optimize /
schedule major changes in technology now
Document
Collection
Scrub
Tokenize
token
physical
M
HashJoin Regex
Left token
GroupBy R
plan
Stop Word token
List
RHS
Count
Word
Count
compute Apache Hadoop, in-memory local mode
“assembler”
code
substrate
…envision GPUs, streaming, etc.
machine
data Splunk, Nagios, Collectd, New Relic, etc.

50. data workflows: SQL
Relational
SQL parser
logical plan,
optimized based on stats
physical plan
query history,
table stats
b-trees, etc.
ERD
table schema
catalog

51. data workflows: SQL vs. JVM
Relational Cascading + Driven
SQL parser SQL-92 compliant parser
(in progress)
logical plan, TODO: logical plan,
optimized based on stats optimized based on stats
physical plan API “plumbing”
query history, app history,
table stats tuple stats
b-trees, etc. distributed compute substrate:
Hadoop, in-memory, etc.
ERD flow diagram
table schema tuple schema
catalog endpoint usage DB

52. Intro to Cascading
Document
Collection
Scrub
Tokenize
token
M
HashJoin Regex
Left token
GroupBy R
Stop Word token
List
RHS
Count
Word
Count
5. tutorial:
for the impatient

53. “Cascading for the Impatient”
cascading.org/category/impatient/
‣ a series of introductory tutorials and code samples
‣ 1:1 code comparisons in Scalding, Cascalog, Pig, Hive
Document
Collection
Scrub
Tokenize
token
M
HashJoin Regex
Left token
GroupBy R
Stop Word token
List
RHS
Count
Word
Count

54. 1: copy
public class
Main
{
public static void
main( String[] args )
{
String inPath = args[ 0 ];
String outPath = args[ 1 ];
Source
Properties props = new Properties();
AppProps.setApplicationJarClass( props, Main.class );
HadoopFlowConnector flowConnector = new HadoopFlowConnector( props );
// create the source tap
Tap inTap = new Hfs( new TextDelimited( true, "t" ), inPath );
M // create the sink tap
Tap outTap = new Hfs( new TextDelimited( true, "t" ), outPath );
Sink
// specify a pipe to connect the taps
Pipe copyPipe = new Pipe( "copy" );
// connect the taps, pipes, etc., into a flow
FlowDef flowDef = FlowDef.flowDef().setName( "copy" )
.addSource( copyPipe, inTap )
.addTailSink( copyPipe, outTap );
// run the flow
flowConnector.connect( flowDef ).complete();
1 mapper }
}
0 reducers
10 lines code

55. wait!
ten lines of code
for a file copy…
seems like a lot.

56. same JAR, any scale…
MegaCorp Enterprise IT:
Pb’s data
1000+ node private cluster
EVP calls you when app fails
runtime: days+
Production Cluster:
Tb’s data
EMR w/ 50 HPC Instances
Ops monitors results
runtime: hours – days
Staging Cluster:
Gb’s data
EMR + 4 Spot Instances
CI shows red or green lights
runtime: minutes – hours
Your Laptop:
Mb’s data
Hadoop standalone mode
passes unit tests, or not
runtime: seconds – minutes

57. 2: word count
Document
Collection
Tokenize
GroupBy
M token Count
R Word
Count
1 mapper
1 reducer
18 lines code gist.github.com/3900702

58. Cascading / Java Document
Collection
M
Tokenize
GroupBy
token Count
String docPath = args[ 0 ]; R Word
String wcPath = args[ 1 ]; Count
Properties properties = new Properties();
AppProps.setApplicationJarClass( properties, Main.class );
HadoopFlowConnector flowConnector = new HadoopFlowConnector( properties );
// create source and sink taps
Tap docTap = new Hfs( new TextDelimited( true, "t" ), docPath );
Tap wcTap = new Hfs( new TextDelimited( true, "t" ), wcPath );
// specify a regex to split "document" text lines into token stream
Fields token = new Fields( "token" );
Fields text = new Fields( "text" );
RegexSplitGenerator splitter = new RegexSplitGenerator( token, "[ [](),.]" );
// only returns "token"
Pipe docPipe = new Each( "token", text, splitter, Fields.RESULTS );
// determine the word counts
Pipe wcPipe = new Pipe( "wc", docPipe );
wcPipe = new GroupBy( wcPipe, token );
wcPipe = new Every( wcPipe, Fields.ALL, new Count(), Fields.ALL );
// connect the taps, pipes, etc., into a flow
FlowDef flowDef = FlowDef.flowDef().setName( "wc" )
.addSource( docPipe, docTap )
.addTailSink( wcPipe, wcTap );
// write a DOT file and run the flow
Flow wcFlow = flowConnector.connect( flowDef );
wcFlow.writeDOT( "dot/wc.dot" );
wcFlow.complete();

59. Scalding / Scala Document
Collection
M
Tokenize
GroupBy
token Count
R Word
Count
// Sujit Pal
// sujitpal.blogspot.com/2012/08/scalding-for-impatient.html
package com.mycompany.impatient
import com.twitter.scalding._
class Part2(args : Args) extends Job(args) {
val input = Tsv(args("input"), ('docId, 'text))
val output = Tsv(args("output"))
input.read.
flatMap('text -> 'word) {
text : String => text.split("""s+""")
}.
groupBy('word) { group => group.size }.
write(output)
}

60. Cascalog / Clojure Document
Collection
M
Tokenize
GroupBy
token Count
R Word
Count
; Paul Lam
; github.com/Quantisan/Impatient
(ns impatient.core
(:use [cascalog.api]
[cascalog.more-taps :only (hfs-delimited)])
(:require [clojure.string :as s]
[cascalog.ops :as c])
(:gen-class))
(defmapcatop split [line]
"reads in a line of string and splits it by regex"
(s/split line #"[[](),.)s]+"))
(defn -main [in out & args]
(?<- (hfs-delimited out)
[?word ?count]
((hfs-delimited in :skip-header? true) _ ?line)
(split ?line :> ?word)
(c/count ?count)))

61. Hive Document
Collection
M
Tokenize
GroupBy
token Count
R Word
Count
-- Steve Severance
-- stackoverflow.com/questions/10039949/word-count-program-in-hive
CREATE TABLE input (line STRING);
LOAD DATA LOCAL INPATH 'input.tsv'
OVERWRITE INTO TABLE input;
SELECT
word, COUNT(*)
FROM input
LATERAL VIEW explode(split(text, ' ')) lTable AS word
GROUP BY word
;

62. Pig Document
Collection
M
Tokenize
GroupBy
token Count
R Word
Count
-- kudos to Dmitriy Ryaboy
docPipe = LOAD '$docPath' USING PigStorage('t', 'tagsource')
AS (doc_id, text);
docPipe = FILTER docPipe BY doc_id != 'doc_id';
-- specify regex to split "document" text lines into token stream
tokenPipe = FOREACH docPipe
GENERATE doc_id, FLATTEN(TOKENIZE(text, ' [](),.')) AS token;
tokenPipe = FILTER tokenPipe BY token MATCHES 'w.*';
-- determine the word counts
tokenGroups = GROUP tokenPipe BY token;
wcPipe = FOREACH tokenGroups
GENERATE group AS token, COUNT(tokenPipe) AS count;
-- output
STORE wcPipe INTO '$wcPath' USING PigStorage('t', 'tagsource');
EXPLAIN -out dot/wc_pig.dot -dot wcPipe;

63. 3: wc + scrub
Document
Collection
Scrub GroupBy
Tokenize
token token
Count
M
R Word
Count
1 mapper
1 reducer
22+10 lines code

64. 4: wc + scrub + stop words
Document
Collection
Scrub
Tokenize
token
M
HashJoin Regex
Left token
GroupBy R
Stop Word token
List
RHS
Count
1 mapper Word
1 reducer Count
28+10 lines code

65. 5: tf-idf
Unique Insert SumBy
D
doc_id 1 doc_id
Document
Collection
M R M R M RHS
Scrub
Tokenize
token
HashJoin
M
RHS
token
HashJoin Regex Unique GroupBy
DF
Left token token token ExprFunc
Count CoGroup
Stop Word tf-idf
List
RHS
M R M R M R
TF-IDF
M
GroupBy
TF
doc_id,
token Count
GroupBy Count
token
M R M R
Word
R M R Count
11 mappers
9 reducers
65+10 lines code

66. 6: tf-idf + tdd
Unique Insert SumBy
D
doc_id 1 doc_id
Document
Collection
RHS
M R M R M
Assert Scrub
Tokenize
token
HashJoin Checkpoint
M
M
RHS
token
HashJoin Regex Unique GroupBy
DF
Left token token token Count ExprFunc
CoGroup
tf-idf
Stop Word
List RHS
M R M R M R
TF-IDF
M
GroupBy
TF
doc_id,
Failure token Count
Traps GroupBy Count
token
M R M R
Word
Count
R M R
12 mappers
9 reducers
76+14 lines code

67. deployed on AWS…
elastic-mapreduce --create --name "TF-IDF"
--jar s3n://temp.cascading.org/impatient/part6.jar
--arg s3n://temp.cascading.org/impatient/rain.txt
--arg s3n://temp.cascading.org/impatient/out/wc
--arg s3n://temp.cascading.org/impatient/en.stop
--arg s3n://temp.cascading.org/impatient/out/tfidf
--arg s3n://temp.cascading.org/impatient/out/trap
--arg s3n://temp.cascading.org/impatient/out/check
aws.amazon.com/elasticmapreduce/

68. results? doc_id tf-idf
doc02 0.9163
token
air
doc05 0.9163 australia
doc05 0.9163 broken
doc04 0.9163 california's
doc04 0.9163 cause
doc02 0.9163 cloudcover
doc04 0.9163 death
doc04 0.9163 deserts
doc03 0.9163 downwind
doc_id text …
doc01 A rain shadow is a dry area on the lee back side of a mountainous area. doc02 0.9163 sinking
doc02 This sinking, dry air produces a rain shadow, or area in the lee of a mountain doc04 0.9163 such
with less rain and cloudcover. doc04 0.9163 valley
doc03 A rain shadow is an area of dry land that lies on the leeward (or downwind) doc05 0.9163 women
side of a mountain. doc03 0.5108 land
doc04 This is known as the rain shadow effect and is the primary cause of leeward doc05 0.5108 land
deserts of mountain ranges, such as California's Death Valley. doc01 0.5108 lee
doc05 Two Women. Secrets. A Broken Land. [DVD Australia] doc02 0.5108 lee
zoink null doc03 0.5108 leeward
doc04 0.5108 leeward
doc01 0.4463 area
doc02 0.2231 area
doc03 0.2231 area
doc01 0.2231 dry
doc02 0.2231 dry
doc03 0.2231 dry
doc02 0.2231 mountain
Unique Insert SumBy
D
doc_id 1 doc_id
Document
Collection
RHS
M R M R M
doc03 0.2231 mountain
Assert Scrub
Tokenize
token
HashJoin Checkpoint
M
M
RHS
token
HashJoin Regex Unique GroupBy
DF
Left token
doc04 0.2231 mountain
token token Count ExprFunc
CoGroup
tf-idf
Stop Word
List RHS
M R M R M R
TF-IDF
GroupBy
M
doc01 0.0000 rain
TF
doc_id,
Failure token Count
Traps GroupBy Count
token
doc02 0.0000 rain
M R M R
Word
Count
R M R
doc03 0.0000 rain
doc04 0.0000 rain
doc01 0.0000 shadow
doc02 0.0000 shadow
doc03 0.0000 shadow
doc04 0.0000 shadow

69. comparisons?
compare similar code in Scalding (Scala) and Cascalog (Clojure):
sujitpal.blogspot.com/2012/08/scalding-for-impatient.html
based on: github.com/twitter/scalding/wiki
github.com/Quantisan/Impatient
based on: github.com/nathanmarz/cascalog/wiki

70. Intro to Cascading
Document
Collection
Scrub
Tokenize
token
M
HashJoin Regex
Left token
GroupBy R
Stop Word token
List
RHS
Count
Word
Count
6. code:
sample apps

71. Social Recommender
filter
Twitter stop words
tweets
calculate
QA
similiarity
threshold
min, max
Neo4j
LDA Redis
github.com/Cascading/SampleRecommender
‣ social recommender based on Twitter: suggest users who tweet about similar stocks
‣ instead of a cross-product (potential bottleneck) this runs in parallel on Hadoop
‣ uses a stop word list to remove common words, offensive phrases, etc.
‣ one tap measures token frequency: for QA, adjust stop words, improve filter, etc.
‣ adapted in Spring by Costin Leau

72. SocRec: architecture
Twitter filter low-freq
firehose source stop words
tweets batch updates
( uid, tweet, t )
checkpoint:
tokenized tweets
calculate checkpoint: analysis +
QA
similiarity token frequency curation
checkpoint: similarity
similar users thresholds
threshold
min, max
sink
sink sink
Neo4j:
social Redis
graph LDA:
topic results
(uid: uidx, rank)
trending

73. SocRec: results
uid recommend weight
carbonfiberxrm ClosingBellNews 0.1459
carbonfiberxrm DJFunkyGrrL 0.0870
ClosingBellNews DJFunkyGrrL 0.1491
CloudStocks DJFunkyGrrL 0.1206
ElmoreNicole DJFunkyGrrL 0.1798
EsNeey alexiolo_ 0.8603
...

74. City of Palo Alto open data
Regex Regex
tree
Scrub
filter parser species
M
HashJoin
Left Geohash
CoPA
GIS exprot Tree
Metadata M
RHS RHS
tree
Regex Checkpoint
road
Regex Regex
tsv
parser tsv filter Tree Filter GroupBy Checkpoint
parser CoGroup
Distance tree_dist tree_name shade
M
R M R M RHS
M
HashJoin Estimate Road
Left Albedo Segments Geohash CoGroup
Road
Metadata GPS
Failure RHS M logs
Traps R
road
Geohash
M
Regex
park
filter reco
M
park
github.com/Cascading/CoPA/wiki
‣ GIS export for parks, roads, trees (unstructured / open data)
‣ log files of personalized/frequented locations in Palo Alto via iPhone GPS tracks
‣ curated metadata, used to enrich the dataset
‣ could extend via mash-up with many available public data APIs
Enterprise-scale app: road albedo + tree species metadata + geospatial indexing
“Find a shady spot on a summer day to walk near downtown and take a call…”

75. CoPA: log events

76. CoPA: results 0.12
Estimated Tree Height (meters)
0.10
0.08
count
0
density
100
0.06 200
300
0.04
0.02
0.00
0 10 20 30 40 50
avg_height
‣ addr: 115 HAWTHORNE AVE
‣ lat/lng: 37.446, -122.168
‣ geohash: 9q9jh0
‣ tree: 413 site 2
‣ species: Liquidambar styraciflua
‣ avg height 23 m
‣ road albedo: 0.12
‣ distance: 10 m
‣ a short walk from my train stop ✔

77. drill-down
blog, code/wiki/gists, jars, list, DevOps products:
cascading.org/
github.org/Cascading/
conjars.org/
goo.gl/KQtUL
concurrentinc.com/
pnathan@concurrentinc.com
@pacoid