Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
1
2
Or retailer…or bank…
3
4
5
6
7
Brand sentiment – By capturing and analyzing customer comments on Facebook, Twitter, and LinkedInin order to improve customer experience and optimize campaign performance.
Predictive maintenance – By analyzing a continuous stream of machine data diagnostics, you canpredict when the performance of machinery is degrading or even worse potentially about to breakdown.
Insider threats – By looking for anomalies hidden in data about user behavior, you identify suspiciousbehavior and pinpoint potentially high-risk employees
Network optimization – By understanding usage patterns, and predicting customer trends, you canoptimize your distribution network
Propensity to churn – Maybe your business suffers from a customer turnover in a highly competitivemarket. How can you determine a customer’s propensity to churn, or in other words, a likelihood theywill leave you as a customer so that you can offer new services or deals in order to keep them.
Fraud detection – Identify purchases or insurance claims that may have a high probability of beingfraudulent by analyzing not only the transactional information, but also electronic documents.
Asset tracking – Track high value assets and identify abnormal behavior that may put assets at risk ofloss, or identify inefficient usage that is costing your business money.
Personalized care – Use advanced analytics to create personalized treatments for patients, such ashow Mitsui Knowledge Industry offers personalized cancer treatment based on genome analysis
Smart Equipment: Predictive maintenance; real-time spatial information; agile logistics (parts, service);smart products; pay-per-use (subscription)Benefits
Reduced maintenance downtime and cost.Prolonged lifetime of machinery and equipment.Fewer / zero failures => increased productivity, business performance and reduced reputationalrisk.
8
Automatically adapt to changing circumstances in the environment.Optimized use, increase security, prevent accidents.Smoother more automated operations with reduced costs.New business models – move from high capital outlay to pay-per-use(subscription) model with value-added services
Smart vending: Predictive maintenance; precision marketing & demand assessment;pricing; planogram updates; consumables managementBenefits
Customized offersSupply chain / fleet management optimizationNo stock-outs
Smart cities: Fleet management; Insurance / pay-as-you drive (how-you-drive); smartparking; fuelling / electric vehicle charging;Benefits
Optimized use, increase security, prevent theft, minimize damage.Smoother more automated fleet operations with reduced costs.Improved driver experience.Tailored services / offers with increased loyalty.Traffic flow, street lighting regulated according to real-time traffic updates.
8
With this explosion of information, the first question asked is, “What can Big Data doto help grow my business?”
As we can see from the analysis of trends and outcomes, Big Data can help informand enable your Strategy, deliver previously unknown Customer Insights, SupportCustomer Buying Behaviors and Develop New Business Models to drive CompetitiveAdvantage
9
http://www.youtube.com/embed/v5kr9hxC8D8/?autoplay=1&rel=0&feature=youtu.be
10
Belgian Railways: You know this…its about traveler experience and uptime withreduced costs. The operational practices applied to purchasing, ordering and goodsmovements. The customer needed to understand The specific parameters that canbe changed and optimized in inventory policies, the Specific topology of the supplychain , Various demand patterns for parts – exhibiting both random and systematiceffects, Special cases like dangerous goods, perishable goods, etc. and dataavailability. SAP Data Science Services solution determines storage compartmentsand optimal volume of parts. Delivering a targeted inventory of parts which allowsfor fast access yet drives to the lowest possible cost against business constraints andstrategic objectives. Freeing up working capital and understanding he tradesbetween inventory level and profitability.
NFL: Use best in class, interactive Fantasy Football experience to turn casual fans intorabid followers. Business and Technical Challenges; Deliver a premier anddifferentiating experience to increase traffic and usage of NFL Fantasy Football siteand to increase viewership/attendance of NFL broadcasts and games through vestedinterest in multiple teams and players across teams. Solution was a PlayerComparison Tool for the NFL Fantasy Football platform This tool will give all fantasyusers the ability to make pro-active decisions with respect to comparing playersregarding their fantasy team including sit/start, add/drop and trade decisions acrossan entire season.
11
Custom-developed algorithms & > 70 models that predict over 90 player-performance statistics across positions (SAP Data Science) with Advancedvisualization and real-time processing of Big Data (SAP Lumira, SAP HANA cloud).Benefits Significant increase in forecast accuracy compared to previous site:Competitive advantage to fantasy football fans with real-time analysis and enhanceduser experience
T-Mobile: They being able to respond to the needs of customers in real time wouldgive us an incredible competitive advantage and improve the quality of the customerexperience. They needed to be able to uncover customer insights and then act onthose insights in minutes, not weeks. Dynamic read outs on the upsell/cross sellperformance of store and call centers. Easy, fast assess to the performance of allcampaigns (e.g. by geo, by store, etc.)Quicker forecast of the financial impact of marketing campaigns.
Alliander: Large Dutch grid operator for delivering energy and gas. Gain insight fromlarge volumes of data (~3.5 billion records per year) generated by over 22,000sensors spread across 400 substations for 35 million customers. Reduce process time(~2 months) and increase the frequency of analysis (more than once a year)Improve forecast accuracy to maintain safe operations and drive investmentdecisions about replacing assets. We developed and refined use cases around theforecasting their peak load and real-time load analysis as well. The issue is that theyhave these grids, overload could impair the longevity and create some sort of outageand in real time being able to analyze the changes and of there is an overload onpiece of equipment we can adjust for what is happening in grid in real-time theoperator can make immediate decisions o transfer and shut down to transfer assets ,providing a forecast.
Pirelli: You know this story, needed to connect with their end customer based onthey had a limited ecommerce channel but mostly dependent on automotive dealers,tire stores and online wholesalers. Based on sensor data could predict and then beproactive in campaigns to reach as a service.
eBay: true Big Data story, PB of data to analyze, 200 analysis at org trying to do thiswork, without predictive capabilities it was taking forever to get answers out of data.One of the things we found through process was noticing weeks later drop in sales ona product and didn’t know why and then through analysis quickly able to realize thatgoogle was changing search algorithms and their items weren’t coming up on top oflist or at all and they weren't finding as quickly or purchasing and they were able to
11
now realize instantaneously something was off and they were able to realize hadsomething to do with google search engine and take actions to manage this
HSE 24: multi channel retailer looking understand customer better. Looking tounderstand customer returns, this is a huge cost to the HSE 24, how can they bettercommunicate with that type of customer to reduce that type of process, what weleverage for this solutions is the AP audience discovery and targeting from CEIsolution and we use predictive for segmentation to understand them better and wework to help them drive a more effective campaign to reduce this type of activity; agood scenario of Hana, Big Data application and predictive analysis
Burberry: Personalized Consumer Experience. The retailer wants to leverage datafrom a customer’s buying history, social media use, fashion trends and othersources to provide personalized service and on-the-spot recommendationsacross the omniverse. An example of Big Data that is very specific to the retailindustry allowing for a retail stores to display on an iPad specific information about
each customer (The retailer wants to leverage data from acustomer’s buying history, social media use, fashiontrends and other sources to provide personalizedservice and on-the-spot recommendations. for the salespeople to have as customers walk into the retail store. A sales person knows whatthe customer has bought in the past, recommendations of items on the floor that areof a similar taste and any other information that would be helpful to the sales personto make the customer experience more personnel and exclusive. This is an exampleof using data in a strategic way to improve the service you provide your customers.There are many other examples you might think of – that if you had real-timeinformation at your fingertips about your customer it would help improve thatexperience for your customer. This is the opportunity with Big Data today combinedwith a real-time big data platform
11
The impact Big Data is having across many different sectors is profound… Here aresome examples. TheBig question is –what are you doing in your organization orcompany to maximize the value of the data you have? Are you leveraging all the datayou can? Are you identifying the valuable insights that can ultimately transform yourbusiness? And are you thinking about real-time and the impact it can have on yourprocesses and people?
12
13
14
Monitor eBay’s marketplace through automated signaldetection powered by predictive analytics100% Accuracy that a signal is positive at 97%confidence
eBay Early Signal Detection System Powered byPredictive Analytics on SAP® HANA
Business ChallengesŸ Increase ability to separate signal from noise toidentify key changes to the health of eBay’smarketplaceŸ Improve predictability and forecast confidence ofeBay’s virtual economyŸ Increase insights into deviations and their causes
15
Technical ChallengesŸ Detect critical signals from 100 PBs of data in eBayEDWŸ Highly manual process because one model does notfit all the metrics hence requires analyst intervention
Key benefitsŸ Automated signal detection system powered bypredictive analytics on SAP HANA selects best modelfor metrics automatically; increases accuracy offorecastsŸ Reliable and scalable system provides real-timeinsights allowing data analysts to focus on strategictasksŸ Decision tree logic and flexibility to adjust scenariosallows eBay to adapt best model for their data
“HANA is valuable in the sense that it accelerates thatspeed to insight. HANA, with in-memory capability, withmulticore, fast, lots of data, all of that coming togetheris how I think analytics is going to work broadly in thefuture.”David Schwarzbach, VP&CFO eBay North America ateBay Inc.“HANA system will free up all the bandwidth right nowinvolved in figuring out what is going. The user just hasto feed in their metric, doesn’t have to really worryabout which algorithm is the best and be able to use
15
the system because it is inherently intelligent andconfigurable.”Gagandeep Bawa, Manager, North America FP&A ateBay Inc.
15
In order to make these use cases real and use Big Data to innovate you must think of the end-to-end solution from acquiringinfinite amount of data, understand what data is needed in real-time vs. what can stay in non real-time storage, whatalgorithms, calculations/analytics/predictions are needed to find the new signals, apply the key insights to the right people(employees, partners, suppliers, customers) in the right processes in an experience that is engaging, visual and right for instantanswers not matter how complex they are.
DeepAnswer complex questions on granular dataPredict the best next action
AccessibleOn any device or to any userSelf service and intuitive interactions
SimpleNo data preparationNo pre-aggregatesNo tuning
Real timeReal-time streams of dataAsk a question, get an immediate anwer
BroadMassive data scalesMany data types
16
17
18
What is the right architecture?
The answer to everything is not Hadoop – yes I am a heretic. But, I was alive whenwe did ISAM. Sometimes I wonder, if the only reason we have noSQL databases isbecause all the folks who new ISAM have retired or are dead.
My learnings over the last 20 years, is that value & “real-time-ness” of theinformation matters. Your end solution will be a combination of technologies.
At SAP, we want to simplify this for you and so we have created the Real-Time DataPlatform with partners such as IBM, HP, Cloudera, Intel, etc….
19
20
SAP HANA In-Memory Platform : Complete picture
ProcessingNLP, Predictive,R-Integration.Spatial processing, ad-hoc OLAP views.Data virtualization.
IngestionReplication, streaming, ETL.Integration, data cleansing.
ConsumptionHTTP(S), OData, XML/A.ODBC, JDBC, ODBO.SQL, MDX.
DevelopmentJavaScript, HTML 5.Any programming language.App/web services.
21
Decision table.
21
Beyond all of the innovation I have just shown, HANA has a series of pluggable“engines” to help take the processing to the data.
• Spatial – use spatially tagged data to calculate (i.e. distribution routes) or displaythe data graphically on a map
• Search – leverage in-memory to find enterprise data truths• Text Mining – create insight from unstructured data• Stored Procedures/Data Models – Put your data in the context of your business
users• Application and UI services – Write HTML 5 based applications directly against
HANA with no third tier.• BFL – create reusable business functions like currency conversions, etc.• Predictive Analysis library – predict your business behavior in real-time• Planning Engine – Do advanced planning in real-time with no processing delays• Rules Engine – Build flexible business rules on enterprise data
You can see that HANA is far more than just a fast database, it’s a complete platformfor enabling a new generation of applications.
Supports any application
22
60% of HANA use cases are outside of the SAP Landscape440+ Startups are developing on HANA
More than a DB – it’s a true platform• Converged OTLP + OLAP• Embedded business logic
22
23
24
25
To facilitate and ensure success along their Big Data implementation path, we havedefined a « Journey to Big Data success » that you can enter at different stagesdepending on your organization’s level of maturity with Big Data.
• If you have identified the business potential, but are still wondering if it makessense for a Big Data project, start with a design thinking “Innovation session”
• If you already have a defined use case, take advantage of our “Big Data Advisory”service that will help establish their Big Data specific strategy, a roadmap and high-level architecture
• Next, is a “Proof of concept” or “Data Discovery & Insights” service that uses anactual data sample to help the customer identify the hidden signals in their data
• With this step-by-step approach we can tailor the service to your specific needs,taking into consideration your level of maturity, while proposing a logical path to asuccessful Big Data project
26
So how to approach customers.The most logical first step is to assess where your customer stands in regards to BigdataTo do that Value Engineering has developed a Big Data Maturity model that you canuse.In a nutshell based on 5 dimensions that you can see on the top, People, ….., we areable to assess the readiness of an organization in regards to Big Data.This is usually lead by Value Engineering and is based on a questionnaire that they filltogether with the customer main stakeholders
27
What drives Innovation
28
29
30
31