Chal allen enges es & & Oppo pportunities es fo rd party data p 3 rd for 3 a partner erships. Mic ha e l Pa c k, CAT T L a b o ra to ry Pa c k’ s Po inte rs o n ho w a g e nc ie s c a n . b e tte r le ve ra g e 3 rd pa rty da ta a nd priva te se c to r re la tio nships.
Data alone isn’t the answe r . • Agencies need: • Policy guidance, • Tools & technologies, • Research & development, and • Thought leadership that helps reduce anxiety and increase big-data capabilities You Data To prevent this scenario: (and your poor Providers staff) Data from Everywhere Big Data Image courtesy of Karl Petty 2
F or E xample : Waze data c an be a fir e hose ! Note: Waze data excludes jams event type • 3 Month Period of 3/17 – 5/17 displayed • 3
Waze Data Bac kgr ound 4
Wor king with Waze Working with the Data Working with the Company • Redundancy • Legal • Feedback loops • Negotiations • Size • Nothing is really free • Credibility and filtering • Increased Coverage • Faster Response • The ability to truly influence route- choice 5
Common 3 rd Party Data Providers and Services • Speeds and travel times • Data Feeds & APIs • Map and data tiles • O/Ds • Trajectory/Trips • Location-Based Services (LBS) • Mapping • Some are working on volumes and turning movements • Much much more coming soon!!! • Not all provide the same type of data, the same format, etc. even for similar data types Po we re d b y: 6
3 rd Party Data can be AWESOME!!! • But… YOU the purchaser can ruin it!!! I mean, really really ruin it. Procurements can go wrong. And you can also get played. Po we re d b y: 7
Don’t make these mistakes • DUAs – You have the power! • Fight for Great Acceptable use • Fight for (and think about) Sharing with partners • Don’t just do what your neighbor did (but ask them) • Look for model DUAs (I-95 CC for probe data) • Sharing back with the provider the way YOU want to share it • (don’t permanently dumb down your data) • Treat your provider as part of your team, NOT a whipping boy • Be open to communication and vendor discussions • Don’t blend “all” of the requirements • Payment terms based on quality and uptime (where applicable) • Stop focusing on how to pay less. Instead, work to try to get more ! Po we re d b y: 8
Inve st in T ools to Make F use d Data E asy to wor k with • Data is only useful when it is • easily accessible, • usable, and • understandable To managers, planners, operations, and ITS applications… 9
T o be e ffe c tive , you ne e d the following: + + = Domain Tools Data Insights Expertise Analysis & Fusion, Statistics, Visualization & Integration
T e c hnic al Capac ity Ne e ds to Inc r e ase (and dive r sify) 11
Inve st in your te c hnic al c apac ity • Don’t just train Transportation Engineers to do this stuff. • Hire other skill-sets and teach them about transportation • Data Journalists / Analysts / Data Scientists • Consultants can do this, too, but…. • Think long-term (don’t hire then fire) • Train staff and transfer knowledge • Partner with Universities (or other similar institutions) • Invest in Research 12
Be war e of Distr ac tions and Hype 13
Buzzwor ds, Shiny Obje c ts, and Pe e r Pr e ssur e • Blockchain • Machine Learning • Artificial Intelligence (AI) • Business Intelligence • The Cloud • Agile • Etc. Know what they mean. Don’t confuse them. Understand their relevance. Don’t think they’ll solve all your problems. 14
Big Data: Savior or Big F at T e ase Expectations Innovation Peak of Inflated Trough of Plateau of Slope of Trigger Expectations Disillusionment Enlightenment Productivity Time
2018 16
T he Cloud (hype , sale s, or savior ?) • The cloud is EXTREMELY cost effective when you do things the way they want you to do them! • Don’t assume the cloud will save you money or improve capabilities • You don’t have to be in the cloud to be effective and innovative • The cloud should not be used for everything • The cloud is not “all or nothing” • Not all clouds are created equally • Virtualization is not the same thing as cloud computing 17
Know your te r ms… 18
Ope n Data vs. Ope n Sour c e (the r e ’s a diffe r e nc e !) • Well-intentioned people confuse open source and open data . • Making institutional investments based on a misunderstanding of terms can have drastic impacts! • Open Source typically applies to software and applications • Open Data applies to DATA 19
Pac k’s Pr e dic tions for the F utur e … • Data isn’t going to get any smaller . • Deploying data collection infrastructure will become increasingly less necessary—even at signals! • Get your (Current) house in order • Or else the latest and greatest thing won’t matter. • You won’t be ready. • Tools (and newer staff should) make some of this easier: • Think of Tableau as the new Excel. • But that means that expectations are going to go up, too! • We need to invest together and pool our resources for data management and analytics. 20
Nex ext s steps T hank you! Michael L. Pack Director, CATT Laboratory PackML@umd.edu 240.676.4060 21
Recommend
More recommend
