From Predictive Models to Production Apps - Inspire 2017
-
Upload
alteryx -
Category
Data & Analytics
-
view
161 -
download
2
Transcript of From Predictive Models to Production Apps - Inspire 2017
FROM PREDICTIVE MODELS TO PRODUCTION APPSPresented by Austin Ogilvie
September 12, 2017
FORWARD-LOOKING STATEMENTS This presentation includes “forward-looking statements” within the meaning of the Private Securities Litigation Reform Act of 1995. These forward-looking statements may be identified by the use of terminology such as “believe,” “may,” “will,” “intend,” “expect,” “plan,” “anticipate,” “estimate,” “potential,” or “continue,” or other comparable terminology. All statements other than statements of historical fact could be deemed forward-looking, including any projections of product availability, growth and financial metrics and any statements regarding product roadmaps, strategies, plans or use cases. Although Alteryx believes that the expectations reflected in any of these forward-looking statements are reasonable, these expectations or any of the forward-looking statements could prove to be incorrect, and actual results or outcomes could differ materially from those projected or assumed in the forward-looking statements. Alteryx’s future financial condition and results of operations, as well as any forward-looking statements, are subject to risks and uncertainties, including but not limited to the factors set forth in Alteryx’s press releases, public statements and/or filings with the Securities and Exchange Commission, especially the “Risk Factors” sections of Alteryx’s Quarterly Report on Form 10-Q. Thesedocuments and others containing important disclosures are available at www.sec.gov or in the “Investors” section of Alteryx’s website at www.alteryx.com. All forward-looking statements are made as of the date of this presentation and Alteryx assumes no obligation to update any such forward-looking statements.
Any unreleased services or features referenced in this or other presentations, press releases or public statements are only intended to outline Alteryx’s general product direction. They are intended for information purposes only, and may not be incorporated into any contract. This is not a commitment to deliver any material, code, or functionality (which may not be released on time or at all) and customers should not rely upon this presentation or any such statements to make purchasing decisions. The development, release,and timing of any features or functionality described for Alteryx’s products remains at the sole discretion of Alteryx.
AGENDA•Quick overview
•Challenges in deploying models
•Solution
•Customers
•Demo
PRESENTER
To watch a recording of this session from Inspire Europe 2017, visit
alteryx.com/inspire-europe-2017-tracks
OVERVIEW
Founded 2013Headquarters in NYC
We help data teamsbuild & deploy apps
You may knowus from
THE PROBLEM
Making data science actionable is challenging & expensive.
• Lack of understanding the benefits
• Lack of trust in effectiveness
• Technical complexity to
operationalize
• No way to measure ROI
MODEL DEPLOYMENT METHODS
Interactive Dashboards Real-time ApplicationsReports
MODEL DEPLOYMENT METHODS
Interactive Dashboards Real-time ApplicationsReports
W H A T I S A P R E D I C T I V E A P P L I C A T I O N ?
Data-Driven Apps
Oscar Health
InsuranceInsurance
UberTransportation &
Logistics
TurboTaxAccounting
http://www.informationweek.com/big-data/big-data-analytics/big-data-success-remains-elusive-study/d/d-id/1318891
DATA SCIENCE VALUE CHAIN
Apps that reach
customers and front-line
employees operationally
are more valuable than
static reports
D A T A S C I E N C E I S A B O U T P R A C T I C A L , R E A L - W O R L D S O L U T I O N S
Carl wants to watch a good movie.
Hey, Carl. Check these out!
E X P L A N A T I O N I S N ’ T A L W A Y S I M P O R T A N T
Sarah builds the algorithm.
Carl would like Frozen because Cindy liked it.
Movie
1
Movie
2
Movie
3
Movie
4
Movie
5
Movie
6
Movie
7
Movie
8
Movie
9
Movie
10
… Movie
17770
User 1 1 2 3
User 2 2 3 3 4 ?
User 3 5 3
User 4 2 3 2 2
User 5 2 3 5 4 2 4
User 6 2
User 7 2 4 2
User 8 3 1 3 4 5 4
User 9 3
User 10 1 2 2
…
User 480189 4 3 3
Carl
Cindy
http://courses.washington.edu/css490/2012.Winter/lecture_slides/08b_collaborative_filtering_1_r1.pdf
E X P L A N A T I O N I S N ’ T A L W A Y S I M P O R T A N T
Sarah builds the algorithm.
>
PROBLEM
Business ProblemEvaluate Available
DataRequest Data
Access from ITRequest Compute Resources from IT Negotiate with IT for
Requested ResourcesWait for Resources to be Provisioned
Install Languages & Tools
Configure Connectivity, Access,
& Security
RAM/CPU Availability, Scalability, Monitoring
Request Network Config Change
Request to Install Another Package
Model BuildingCompose a
Powerpoint to Share Results
Edit Team Wiki to Document Your
Work
Negotiate with Product on Model Deployment
Timeline
Wait for Engineering to Implement the Model
Test Newly Implemented Model to
Ensure Valid Results
Request Modifications to the Model due to
Unexpected Results
Release the Model to Production
Document Release Notes and
Deployment Steps
Prepare for Change Management
DATA SCIENCE TEAMS FACE A MYRIAD OF CHALLENGES AT EVERY STEP OF THE WAY
Deployments take 12-20 weeks
Cost to deploy 1 model runs in excess of $250,000
< 10% of models make it to production
Your ApplicationsData Scientists Developers
Your Customers
Write more custom code to integrate Customer benefitsPainstakingly rewrite models into other languagesBuild a model in R or Python
SOLUTION
SOLUTION
Business ProblemEvaluate Available
DataRequest Data
Access from ITRequest Compute Resources from IT Negotiate with IT for
Requested ResourcesWait for Resources to be Provisioned
Configure Connectivity, Access,
& Security
RAM/CPU Availability, Scalability, Monitoring
Request Network Config Change
Request to Install Another Package
Model BuildingCompose a
Powerpoint to Share Results
Edit Team Wiki to Document Your
Work
Negotiate with Product on Model Deployment
Timeline
Wait for Engineering to Implement the Model
Test Newly Implemented Model to
Ensure Valid Results
Request Modifications to the Model due to
Unexpected Results
Release the Model to Production
Document Release Notes and
Deployment Steps
Prepare for Change Management
Install Languages & Tools
DATA SCIENTISTS NEED A WAY TO MANAGE THEIR PROJECTS FROM END-TO-END
•Build models in Python and R
• Instant model APIs
•Manage models
•Scale without IT
PROMOTE
ALTERYX + PROMOTE
CUSTOMERS
F e r r a t u m B a n k u s e s S c i e n c e O p s t o
m a k e r e a l - t i m e c r e d i t d e c i s i o n s .
• C r e d i t S c o r e s
• F r a u d C h e c k s
• K Y C
• L i n e A s s i g n m e n t
• R i s k - b a s e d P r i c i n g
• Propensity-to-buy
• Cross- and upsell opportunities
Tendril helps solar, smart thermostat, and other
energy providers target the right homeownersc
Results
• 4x faster time to market• Self-sufficient analytics team• $350,000 saved
DEMO
•Yhat joined Alteryx June 2017
•Easily productionize models
•Model management, made easy
WRAP UP
#inspire16#
alteryx.com/trial
Ready to bring these incredible and tangible benefits to your organization?
Download a FREE Trial of Alteryx and start making your data work for you, instead of you working for your data