Sitting
at my computer is my daily opportunity to collect my thoughts. It's also my opportunity
to scatter my thoughts to the wind.
The
bridge scandal shows Christie's office took Nixonian glee in dirty tricks.
Tricky Dick would be so proud of Crafty Chris.
I
see there will be a new theatre complex at rebuilt Springfield Mall. Good. Just
in time as streaming demolishes brick&mortar cinema.
"What
Is A/B Testing And Why It Is Essential For Your Online Campaigns" (http://bit.ly/1703vdO )
JK--Rapid trial&error refinement.
WashPost
article about privacy issues surrounding in-store tracking of smartphones in mobile
commerce. Industry best practices sorely needed
Getting
a good soaking of January rain. Expected warmup to 60 today. Cold please take
your time returning. Don't need the ice.
Being
requested to do some #bigdata
events in DC later in the month. Always happy to go downtown. But love my
suburban existence.
Facebook
annoyingly asks you to review every place you've ever mentioned being at. As if
we still remember or care.
"Convergence"
is a handy catch-all theme in high-tech. It simply calls attention to a point
on the horizon toward which all paradigms vanish
New
#IBM jk
blogpost: "Koby's predictions for the Internet of Things in 2014 and
beyond" (http://ow.ly/steSi )
#bigdatabytes
#IBM chief
privacy officer Christina Peters can morph her video on demand to mask how she
looks & where she is...wow!!!!
"Physics
of Neuroscience" (http://ow.ly/ssLNl ) JK--Ephaptic coupling,
role of biophotonics in intercellular communication, microtubules
"Top
5 threats 2 IoT devices" (http://ow.ly/ssKKr ) JK--Vulnerabilities in
in-car Wifi, mobile med devices, wrbls, retail inv mon, drones
"Comparing
machine learning classifiers bsd on hyperplanes or decision boundaries" (http://ow.ly/ssJRk )
JK--Over my head but mesmerizing
"13
Awkward Things Youll Only Understand If U Work In Extremely Quiet Office"
(http://ow.ly/srwTG )
JK--Need neutral ambient office noise
If
you believe Christie knew absolutely nothing about this, I'd like to sell you
the George Washington Bridge
Drafted
all of next week's quick-hits. Have warmed up my frozen head for new year. No
radically new topics. Just deepening of some old ones
RT
@kexpplaylist #kexp Stop
Draggin' My Heart Around by Shearwater with Sharon Van Etten JK--This cover
better than Nicks/Petty original.
"By
2018, cars will be self-aware" (http://ow.ly/sqKP3 )
JK--Wasn't that plot of some Stephen King book? Now I'm scared to turn
ignition.
"Th
Limits of Phenomenology" (http://ow.ly/sqFr3 ) JK--#BigData's
limits: "theory is even more essential in understanding complex systems
At
World Trade Center
Retweeted
by jameskobielus
Curious
whether there's any correlation between influence & influenza. Are the most
social people also most likely to spread and/or catch?
RT
@kexpplaylist #kexp Me &
Julio Down by the Schoolyard by Paul Simon from his 1st solo LP JK--Feels like
Paul at hootenanny in the Village
"#IBM #Watson Group
Unveils Cloud-Delivered ...Svcs to ..." (http://ow.ly/sq7s2 )
JK--IBM Watson Discovery Advisor, Analytics, & Explorer
"#IBM Forms New #Watson Group to
Meet Growing Demand 4 Cognitive Innovations" (http://ow.ly/sq6wA )
JK--Will invest >$1B to dev apps+svcs
"New
Form o Spacecraft Proplsion Proposd 4 Uranus Mission" (http://ow.ly/sq4ac )
JK--Elctrc sail powerd by solar wind protons not photons
"Pinterest,
Yahoo, Dropbox & (kind of) quiet content-as-data revolution" (http://ow.ly/sq3bq )
JK--Machine learning on images + text
Species
chauvinism. Biologists called our order "primates" & our species
"sapiens." Essentially, first among animals + only one with wisdom.
#IEEE Rockshtars o
#BigData
Presos" (http://ow.ly/sq1CV ) JK--"Information,
once made digital, can never B called back." Grady Booch #IBM
"#BigData
Shrinks to Grow" (http://ow.ly/spZKJ ) JK--Sez industry
retrenchment nigh. But also increasing verticalization & services focus
"The
World's Most Unusual Data Centers" (http://ow.ly/spYym )
JK--Cool infographic. One of them claims to be "nuke-proof."
"Sustainable
brands & #bigdata
set to go mainstream in 2014" (http://ow.ly/spXSJ )
JK--"Conscious brands" use BD to monitor global impact
"Neural
code-breaker" (http://ow.ly/spWzf ) JK--Makes case for
Turing as pioneering neuroscience use of Bayesian & sequential analysis
CES,
happening the first week of January every year, feels like the tail end of the
Christmas season. Adults unwrap and admire new toys.
Only
good thing re whole sorry "Dennis Rodman in N.Korea" saga is that
nobody respected Rodman to start with. Just bigger laughingstock now
Christie's
bipartisan bonafides starting to fray. I suspect he'll finesse his way out of
this firestorm, though. Test of political bonafides
Polar
vortex is a rotating low-pressure system. So's a hurricane. Will we start 2
refer 2 latter as a "subtropical vortex" 2 B consistent?
Early
2014 shows broad signs of economic improvement. But my tweeting this jinxed all
that. Disregard those encouraging indicators. So sorry
Bought/downloaded
an album from 2001: Neil Halstead "Sleeping on Roads." Wistful warm
hushed folkish stylings reminiscent of Nick Drake.
RT
@kexpplaylist #kexp I'm Waiting
for the Man by David Bowie from Bowie
at the Beeb: Best of BBC Radio Sessions 68-72 JK--Gr8 vsn o VU/Lou
@EmUprichard Ah, yes, grasshopper,
using #IBM #SPSS to analyze #birthcohorts,
#UnderstandingSociety
means that U are indeed a #datascientist
A9:
#bigdatamgmt
Assumption is that "fancy" = "expensive" + "long
learning curve." But viz-intensiv, cost-eff self-svc tools now predominate
A9:
#bigdatamgmt
Notion of "fancy" power tools is become quaint in the democratized
analytic world. More knowledge workers have these tools
A8:
#bigdatamgmt
#DataScience
is about problem solving thru exploratioin of empirical data. Any scientific
field with quant = good bakground
A8:
#bigdatamgmt
There's nothing wrong with getting an advanced degree in stat. But doesn't
necessarily mean you're a critical thinker.
A8
PhD degree leads to false positives.Phd's spend yrs focused on small
problem,may not be suited to your problems #bigdatamgmt
Retweeted
by jameskobielus
A8:
How about a PhD in research methodology... how to set up your data, use
reliable/valid metrics... #bigdatamgmt
Retweeted
by jameskobielus
Q8:
I think this question is too generic, Ph.D or no Ph.D, real life experience
with data stats and you could be an expert #bigdatamgmt
Retweeted
by jameskobielus
A8:
#bigdatamgmt
Most real-world data scientists combine quant skills with domain expertise.
Stat degree doesn't give you the latter.
A8
#datascientist
work varies by industry/function, from biz analyst + data mining to real
science at Google, research #BigDataMgmt
Retweeted
by jameskobielus
A8:
#bigdatamgmt
If PhD were necessary+ sufficient condition 4 #datascience,
#bigdata
revolution would grind 2 halt. Few o them 2 go around
A7:
#bigdatamgmt
To extent that you, as an analyst, have acquired stat analysis/modeling skills
& access #bigdata,
you're a "data scientist"
#bigdatamgmt
A7: #DataScience
is key to doing inquiry, discovery, analysis, & understanding. But there
are many other requirements
Retweeted
by jameskobielus
A7
Working w/BigData >many skills, Data Science just one. #DataScientists
too expense to wrangle data,tune clusters #BigDataMgmt
Retweeted
by jameskobielus
A7:
#bigdatamgmt
Any user of #IBM
#Watson is
"working with #bigdata,"
but the complexities tucked away behind sophisticated/simple interface
A7:
#bigdatamgmt
Any developer, analyst, or user--not just data scientists--can work with #bigdata, as
long as you have suitable tools.
A6
Today Hadoop #MapReduce
is batch, but #YARN
giving some workload management to Hadoop elements #BigDataMgmt
Retweeted
by jameskobielus
A6:
#bigdatamgmt
Bear in mind that in #Hadoop's
unstruct/ETL role, batch is usually sufficient. Real-time handled downstream
via in-mem etc
A6:
#bigdatamgmt
Hadoop's core modeling abstraction, #MapReduce,
runs over real-time stream computng plats such as #IBM InfoSphere
Streams.
#bigdatamgmt
A6: #Hadoop
is batch-oriented, but becoming more interactive with new SQL-on-Hadoop tools
& technologies (eg, @Cloudera
Impala)
Retweeted
by jameskobielus
A6:
#bigdatamgmt
If by #Hadoop
U mean #MapReduce
over #HDFS,
yes, primarily 4 batch. But #MapReduce
over R-Ts (Hbase Cassandra etc) too
A5:
#bigdatamgmt
#Hadoop
itself evolving in ways that blur distinctions btwn it & other #bigdata platforms.
#YARN makes it
more genl purpose
A5:
#bigdatamgmt
#Hadoop's
key/growing role in the #bigdata
universe can't be denied. It's the core unstructured data/analysis platform
A5:
#bigdatamgmt
#BigData
refers to the growing range of data plats that scale along any/all of the
"3 Vs." Hybrid deploys becoming common
A5:
#bigdatamgmt
Only most shortsighted #Hadoop
pure-play vendors/devels believe that. MPP/RDBMS, stream cmptg, #NoSQL, in-mem
also #bigdata
A4:
#bigdatamgmt
If you look at other hot #bigdata
analytic apps (e.g geospatial, mobile, #CXO, IoT), many
are customer/user focused.
A4
Funnel Conversion,Cust Seg,Behavioral,Churn,etc.get lots of press,but prpmise
of Bigdata (hopefully) is a better world #BigDataMgmt
Retweeted
by jameskobielus
A4:
#bigdatamgmt
#BigData
(aka VLDB, stat/pred model), even pre-#Hadoop, often
focused on cust data/analysis (eg NPS). Sentiment = mktg gold
A4
Misconception/oversimplification that #BigData is
only about customers and selling things. #BigDataMgmt
Retweeted
by jameskobielus
A4:
#bigdatamgmt
Not primary, but hot. Current #bigdata boom
triggered in part by CMOs flocking to new ways 2 unlock social-media sentiment
A3:
#bigdatamgmt
In popular mind, #bigdata
=t unstructured, cuz that's bulk o wot's "out there." But that
ignores fact structured often core
A3:
No. Ultimately it's all structured data. Unstructured data is transformed to
structured data to make sense of it. #bigdatamgmt
Retweeted
by jameskobielus
A3:
#bigdatamgmt
Fact is #bigdata
is about advanced analytics on ANY data at ANY scale. Nothing sacrosanct,
value-wise, about unstructured
A2:
#bigdatamgmt
Whole notion of any scale of data "trumping" any other is misguided.
Decis support (ie endgame) needs choice data, not all
A2:
#bigdatamgmt
Sandbox data,4 exploration, usually smaller than operatnl data. But sandbox
where data sci finds patterns (ie core value)
A2:
#bigdatamgmt
See this article by me for thoughts on big+small data joined at hip: http://bit.ly/WQSAys
#bigdatamgmt
A2: #BigData
allows for whole-population #analytics!
See @jameskobielus comments at:
http://linkd.in/19dY2Ul
Retweeted
by jameskobielus
A2
Data doesn’t always speak for itself.Detail can be noisy/dirty.OTOH, aggr'd
data can lead to Simpson's Paradox #BigDataMgmt
Retweeted
by jameskobielus
A2
BigData and small data are relative terms.But both are needed for #analytics #BigDataMgmt
Retweeted
by jameskobielus
A2:
#bigdatamgmt
#BigData is
about seeing diff value from analytics as you ramp up "3Vs." But much
analytic app value (eg BI) @ lower scale
#bigdatamgmt
A1: I now promote my new definition of #BigData =
"Everything, quantified & tracked", which encompasses all domains
Retweeted
by jameskobielus
A1:
#bigdatamgmt
In point of fact, most #bigdata
deploys start at lower scales (eg., 10s of TBs), with no rush to go
"massive" for own sake
A1:
#bigdatamgmt
Popular fixation on "massive volume" distracts from fact that #bigdata is all
about headroom to grow from when need arises
A1:
#bigdatamgmt
My working definition of #bigdata is
"advanced analytics on trustworthy data at all scales." That's a
value-focused POV
A1:
#bigdatamgmt
Well U have th whole "3 Vs" thing, only 1 o which is
"volume." Points 2 fact that #bigdata =
analytics @ scale speed scope
@IBMbigdata Muchas gracias. Me gusta
etweetar en esta tweetchat. Data grande is muy delicioso, pero hay muchos
mitologios sobre lo!
My
inner Youngman: Guy sez: "speak of the devil..." Other guy:
"leave my wife out of this."
#Bigdatamgmt
chat starts in 34 mins! Debunking #BigData Myths
w/ @kirkdborne & @neilraden et moi http://ibm.co/1dRfhKR
RT
@kexpplaylist #kexp Golden Years
by David Bowie from Station to Station JK--Fun fact: Bowie didnt stay w/Angela into their golden
years
Somehow,
ideologically driven individuals have managed to demonize one of the most
neutral/boring/ancient terms in IT lexicon: metadata.
"NoSQL,
Part 1: Coming o Age" (http://ow.ly/snKWr ) JK--And bursting its
britches: ~ 150 NoSQL databases listed at http://NoSQL-databases.org
"Big
brains small minds: Troubling vision o SilVal elites" (http://ow.ly/snKsM )
JK--Hm. Something tells me Cringely'll B flamed severely
"SemWeb
Business: Going Nowhere Slowly" (http://ow.ly/snJQF )
JK--Hmm. In one o nxt wk's q-hs, I outline its potential in cognitive cmptg
RT
@kexpplaylist #kexp Somebody To
Love by Queen frm Day @ Th Races JK--LP also had "My Best Friend."
Shoulda calld LP "Surrealistic Pillow"
"Sexy
Data Science is Team Sport" (http://ow.ly/snH5g )
JK--It's not an orgy. Get your minds out of the gutter. Stay professional,
folks!
Before
long, BYOD (bring yr own device) will morph into BYOT (bring yr own thing).
It's yr thing do whatcha wanna do I cant tellUwho2sockit2
RT
@kexpplaylist #kexp A Change Is
Gonna Come by The Gits from Enter: The Conquering Chicken JK--Rock vsn o Sam
Cooke classic. Etheridge-ish
RT
@kexpplaylist #kexp Walk Right
Back by The Everly Brothers JK--Every tight vocal harmony in rock echoes these
siblings' pioneering sound
RT
@kexpplaylist #kexp Take Pills
by Panda Bear from Person Pitch JK--2007. I like this EP's atmospherics--i.e,
blatant use o space & reverb
Get-rich-quick
commercials where they fan out cashwad as if were playing cards. Reminds of
magician. As if saying "it's a trick--no $ 4 U"
"Still
More R and Python" (http://bit.ly/1acY5i3 ) JK--Steve Miller
seems to be pointing to convergence between the languages over time.
"IoT
Wildly Insecure And Often Unpatchable" (http://wrd.cm/KvIEHB )
JK--Provocative opinion from Schneier. See mine: http://ibm.co/1cNWwTL
BTW,
low in WashDC on 1-20-85 was -4F (http://ow.ly/slOji ). Hell
didnt freez over. Tho w/Reagan starting 2nd term, it sure felt that way
RT
@kexpplaylist #kexp That's Not
Really Funny by Eels frm Souljacker JK--2001. Righteously manic mix. Feels like
inmates partying in asylum
Drafted
my latest #IBM
jk blog: "Koby's Predictions for Internet of Things in 2014 and
Beyond"
5
degrees. Feels like -10, my app tells me. Because we've braced for it, it
doesn't actually feel that cold. It simply feels like January.
Polar
vortex. Superstorm. Snowmaggeddon. What's with the mania for inventing new
weather hype words? God's just dishing out the usual drama
If
you want a 60s US band that was jangly proto power-pop a la Big Star, give a
listen to SF's Beau Brummels (http://ow.ly/skrtI )
New
#IBM jk podcast:
"Foretelling 2014 Trends in #BigData, #Hadoop, Data
Science & More" (http://ow.ly/skp7a ). Yeah, that's my voice.
"Davenport : #bigdata too
imprtnt 2 B left 2 'quants'" (http://ow.ly/sk0qu )
JK--Proposes "narrative" aproch 4 communic8 analytics results
"#Hadoop
Interview Questions" (http://ow.ly/sjZCO ) JK--Good luck getting
Hadoop to sit still for an interview. Elephants are squirmy.
A9:
#CXO Boost
loyalty + control costs by consolidate cust data into high-perf #bigdata
platform w/in-db analytics & solution accelerators
A8:
#CXO Empower
employees with ability to use judgment to addr each situational
cust-interaction instance--& log their actions in #bigdata
A8:
#CXO Empower
employees with #CEM
churn mitigation & #nextbestaction
apps powered by best customer data & best-fit propensity models.
A8:
#CXO Empower
empls with access to complete customer profile, history, & interactions,
contextualized for task (mktg, svc, etc) at hand
Retweeted
by jameskobielus
RT
@businessgp: #cxo A7 You can't
measure loyalty in real time, you only measure actions & interactions.
Loyalty is time dependant.
Retweeted
by jameskobielus
A7:
#CXO Assessing
customer loyalty in real-time is a bit like measuring whether marriage will
endure based on how loudly they argue.
A7:
#CXO In
real-time, you don't "gauge" customer loyalty. Rather, you measure
whether stream of interactions is consistent with loyalty
A7:
#CXO There's no
loyalty to be measured in real-time. All you can measure in real-time is
mercurial: attention, intention, & intensity.
A6:
#CXO Potential
customers who ask the most questions, then sign up, are often building
downstream case for staying with you.
A6:
#CXO Prospective
customers who've invested time listening to pitch are more likely to sign up.
If continue to spend time, likely to stay
A6:
#CXO Customers
whose friends, families, and coworkers all use your products are less likely to
churn. No one wants to be an outlier.
A6:
#CXO The ones
that have made the deepest investment in your products are likely to stay.
Cut-and-run is wrenching/costly for them
A5:
#CXO Customers
growing disloyal are more likely to mention competitors favorably in same
breath in which they're complaining about you
A5:
#CXO Customers
whose loyalty is fraying are often those who call cust svc with the same
complaint over and over. You're not listening
A5:
#CXO Customers
whose loyalty is waning are less likely to log into your customer portal,
purchase anything, respond to direct mail, etc
A4:
#CXO #BigData is a
barrier to loyalty when you develop a data fetish and imagine that volume,
velocity, & variety = value to customer
A4.
when you think all data has meaning and you ask the wrong questions...remember
correlation doesn't mean causality #CXO
Retweeted
by jameskobielus
RT
@Hoovers: A4: Big Data can become a
barrier to building customer loyalty when data collection is perceived as
violating privacy. #CXO
Retweeted
by jameskobielus
#cxo A4: When you
get hyperfocused on the data collection & forget to build the data science
around it.
Retweeted
by jameskobielus
A4:
#CXO #BigData
becomes barrier to loyalty when keep it in non-oper systems--e.g, archives
& sandboxes--but don't leverage it in channels
A4:
#CXO #BigData
becomes a barrier to loyalty when you faii to mine it for customer
segmentation, personalization, & experiential patterns
A4:
#CXO #BigData
becomes barrier to building #loyalty when
you focus on it to exclusion of simple human interactions in channels
A3:
#CXO #BigData
analytics drive next-best action apps that your channels use to inform
satisfaction-enhancing interaction w/customers
A3:
#CXO #BigData is
repository of historical data for building predictive models of customer
loyalty/churn. Helps you assess path forward
A3:
#CXO #BigData
360-view depicts how well--or poorly--you've served your customers. Mine it to
find where you've succeeded vs. screwed up
A2:
#CXO Customer
happiness is key to loyalty/profitability: far more expensive to win back a
churned customer than to delight a kept one
A2:
#CXO Just as
important as listening, you should constantly offer customer unique,
personalized value based on what you're hearing
A2:
#CXO First step
is to always be listening to your customer thru every channel. If they know
you're tuned in, loyalty often follows
A1.
when your customer becomes your salesforce (referrals) #CXO
Retweeted
by jameskobielus
A1:
#CXO Customer
loyalty metric is when they come to you first & often with their new
requirements. They know you're best suited to respond
A1:
#CXO Customer
loyalty tested when attractive alternatives emerge but customer remains. You
offer some unique value, perhaps experience
A1:
#CXO Best
measures of customer loyalty include repeat purchases, steady renewals,
recommendations to friends, & high satisfaction scores
@IBMbigdata Pleasure to be on this #CXO chat. The
first of the new year!
@IBMbigdata Cracking my knuckles,
getting ready for #CEM
#CXO chat in one
minute. #voc #nps #custserv
Join
#CXO chat at
12noon (EST) today. "The Matter of Loyalty" (http://on.fb.me/1i1JKGT )
#ux #cx #nps. I'll be on it
w/ @maldyj et al
Headline
says former FCC Chair Julius Genachowski has joined Th Carlyle Group. Ex-govt
officials joining DC-based Carlyle has become cliche
RT
@kexpplaylist #kexp A Ho A Hand
by Famy JK--2014. French group living in Britain . Great one. I loved this
song from the very first note.
"#BigData
Analytix Can Delivr US Energy Independnce" (http://ow.ly/sj1xs )
JK--Geo analytics + fracking. Shock absorbers from geopolitics
"Data
Mining Techniques" (http://ow.ly/siZ6J ) JK--Regression,
clustering, & classification analysis; outlier detection; assoc rule lrnng
"Graph
Approach for Data Mining in BI Applications" (http://ow.ly/siYsm )
JK--Good overview of graph analysis with visualizations.
#Data scientists:
IT's new rockshtars" (http://ow.ly/siXWa ) JK--Other IT pros want
to rock, roll, & reboot careers around data science.
"What
to expect of Internet of Things in 2014" (http://ow.ly/siXsy )
JK--"Most consumers don't know IoT exists." Embedded, pervasive.
We're
OK in NoVa right now, but temp expected to drop 40 degrees today into the
overnight. Just happy that cold alone doesn't need shoveling
Glad
we had warmer temps & rain today. Loosened the ice so I could scrape it
from driveway. Deeper freeze coming. Don't want that freezing.
RT
@kexpplaylist #kexp 1880 or So
by Television JK--1992. Soulful highlight from Tom Verlaine & crew's brief
return to recording.
I
like to pronounce "rockstar" as "rockshtar" and do it in an
accent that implies I'm from Texarkana .
No reason.
Google
"Washington Post Scott Adams willpower" Voila! (http://m.washingtonpost.com/business/read-this-if-you-want-to-be-happy-in-2014/2014/01/02/d96370f0-7192-11e3-9389-09ef9944065e_story.html …)
When
I tweet a reference to a specific theme, article, author, & pub, that's
ample fodder for those who wish to Google for the URL.
Serena
Altschul interviewing James Carville on "CBS Sunday Morning." In Oct
2012 I sat next to him on Delta Shuttle. Gangly, gaunt, skeletal
Good
WashPost article by Scott Adams on need to substitute knowledge for willpower
and need for systems of knowledge to guide right action.
RT
@kexpplaylist #kexp You Are My
Sunshine / Open Up Your Heart by Anne Murray from There's a Hippo In My Tub
JK--1977. Gr8 version.
No
matter how deeply it dips below zero, I retain sense memory of having survived
winter after winter of this in the Midwest .
People adapt.
Watched
1st episode of "Mad Men." Some good dialogue amid the smoking &
drinking & workplace sexual harassment. Hard 2 feel nostalgic 4 1962
RT
@kexpplaylist #kexp Avant
Gardener by Courtney Barnett JK--2013. She brilliantly calls her act
"Courtney Barnett & the Courtney Barnetts"
Today
been below freezing all day. But doesnt feel uncomfortbly cold. I'm acclimated.
Now all I need is 2 hibernate w/some pic-a-nic baskets
Country-western
& Stetsons. Only musical genre with mandatory headwear for men. What is
this: the NFL?
I'm
only 6 paragraphs into Billy Crystal's memoir and I've not stopped laughing.
Perfect weekend reading on the sofa.
Got
a great business idea: drive-in theatre that only shows short Hollywood
promo films. Call it "Trailer Park." You in?
RIP
Phil Everly. One half of the one of the purest sibling vocal harmony duos in
pop music history. And a fine guitarist.
Getting
addicted to "Louie." Now into season 3. Louis C.K. seems like a tired
old man. But he's 9 years my junior. That fact gives me pause.
