An Introduction To Advanced Analytics - RapidMiner

Transcription

An Introduction SINESS INTELLIGENCEADVANCEDANALYTICS

An Introduction toAdvanced Analytics.Where Business Intelligence Systems End.and Predictive Analytics Tools BeginAdvanced Analytics is “the analysis of all kinds of data using sophisticated quantitative methods (for example,statistics, descriptive and predictive data mining, simulation and optimization) to produce insights that traditionalapproaches to business intelligence (BI) — such as query and reporting — are unlikely to discover.” 11Gartner, Magic Quadrant for Advanced Analytics Platforms, Gareth Herschel Alexander Linden Lisa Kart, 19 February 2014Figure s veModelingPredictiveAnalyticsOptimization& SimulationText AnalyticsMultimediaAnalyticsData InfrastructureRDBS, Hadoop, Text Indexing, NoSQL, FilesStructuredtablesDataSemi-structuredXML, graphs, seriesUnstructuredtexts, images, audio, videoPage: 1

AnalyticsAnalytics refers to the skills, technologies, applications and practices for continuousiterative exploration and investigation of data to gain insight and drive businessAnalytics.1. Business Intelligence -- traditionally focuses on using a consistent set ofmetrics to measure past performance and guide business planning. BusinessIntelligence consists of querying, reporting, OLAP (online analytical processing),often.”2. Advanced Analytics -- goes beyond Business Intelligence by using sophisticatedBusinessIntelligencedata into meaningful and useful information for business purposes. BI tools, at themost basic level, help business users interpret voluminous data. BI focuses on thestorage and retrieval of data from the past, using technologies like data cubes andquery engines. Being able to measure past performance is essential in complexmarket advantage.The term Business Intelligence also often refers to the creation and maintenancequerying and reporting. Besides the querying and visualization of data, traditionalBI environments make it possible to implement rule-based alerts to inform decisionAdvancedAnalyticsgoes beyond Business Intelligence and expands the horizons of traditional analytics.With traditional analytics, analysts look at the data and ask “What happened?” WithMethodologies and technologies from both statistics and computer science haveplayed an important role in the development of advanced analytics, and havecontributed to the discipline of Advanced Analytics. The main contributions comefrom Machine Learning and Data Mining.Page: 2

MachineLearningData MiningData Mining has enriched machine learning by also covering the necessary steps ofdata integration and preprocessing in order to create better models. Machine learningfocuses on model building. Today, the term data mining is used less often. Amonganalysts, this term has mainly been replaced by Predictive Analytics, DescriptivePredictiveAnalyticsPredictive analytics is the practice of analyzing data to make statistically accuratepredictions about future events. Predictive Analytics encompasses a variety oftechniques from computer-aided statistics, machine learning, and data mining thatexploit patterns found in historical and transactional data in order to extrapolate tofuture events, and, by that means, predict the most likely future. Models describingthose patterns capture relationships among many more factors than human beingsopportunities.Generally, the term predictive analytics is used to mean predictive modeling, that is,disciplines, such as descriptive modeling and decision modeling or optimization. Thesemarket segment.Going BeyondTabular DataWhen most business people think about data, they envision a classic database, or a data like text collections.unstructured data like images, audio, or video into a structured formatthat can be used as the base for predictive or descriptive analytics.Analytics models that can process unstructured data provide better predictions.Page: 3

Intelligence and Advanced AnalyticsQuick Comparison TableBusiness IntelligenceAdvanced AnalyticsOrientationRearviewFutureTypes of questionsWhat happenedWhen, who, how manyWhat will happen?What will happen if we change thisone thing? What’s next?Reporting (KPIs, metrics)Predictive ModelingAutomated Monitoring/Alerting(thresholds)Data MiningDashboardsMultimedia MiningScorecardsDescriptive ModelingOLAP (Cubes, Slice & Dice, Drilling)Statistical / Quantitative AnalysisAd hoc querySimulation & OptimizationBig DataYesYesData typesStructured, some unstructuredStructured and UnstructuredKnowledge GenerationManualAutomaticUsersBusiness UsersData scientists, Business analysts,IT, Business UsersBusiness InitiativesReactiveProactiveMethodsText MiningPage: 4

Advanced Analytics vs. Business IntelligenceSample QuestionsWhere Business Intelligence is focused on reporting and querying, Advanced Analytics is about optimizing, correlating,and predicting the next best action or the next most likely action. This table compares the types of questions.VerticalApplicationBusiness IntelligenceAdvanced AnalyticsFinancial ServicesFraud detectionWhich credit cardtransactions have beenHow likely is it that acredit card transactionwill be fraudulent?Retail and ConsumerProductsTelecommunications andMediaInvestment holdingcompany: BusinessdevelopmentWhich companies wentbankrupt last year?How likely is it thata company will gobankrupt?Insurance UnderwritingdepartmentProvide me with a listof people who hadaccidents in the last 6months.Is this person likely tohave an accident andshould we continue toinsure them?RetailerWho bought both beerand pretzels?How likely am I topurchase a particularproduct if I havepurchased anotherproduct? (If I buy beer,how likely am I to buypretzels?)Sales and marketingWhat books did you buyfrom our website lastyear?What books might yoube most interested inbased on your previouspatterns of interest?CPG Sales and marketingWhat is the amount oftime between purchasesof our product by asingle customer?Predict when individualconsumers will exhausttheir supply of a certainproduct.Telco Marketing,customer serviceWhich customerscancelled their service?Which customers will bemost likely to leave yourservice?Which ads were moreMedia marketingManufacturingContinued onto the next pageWhich ads will be morechannel, on a particularnight?channel?Media Sales andmarketingWhat movies did yourcustomers rent orpurchase?What movie will yourcustomers be most likelyto buy or rent?Product design anddevelopmentWhat ingredients can befound in our products?Which chemicalformulations will createattributes?Page: 5

Continued from page 6VerticalCross industryApplicationBusiness IntelligenceAdvanced AnalyticsProduct developmentWhat happened last timewe introduced a newproduct?What will be the impactmaintenanceWhat’s the last time thismachine broke down?What’s the maximumproduction we haveever realized from thismachine?If I changed themaintenance schedulefor this machine, howwould it impact myproduction throughput?Customer serviceWhat reasons didcustomers give forcancelling their service?What actions will causecustomers to leave yourservice?Economic developmentWhere are our biggestmarkets currently?Where should we focusWhich partner wasresponsible for the mostWhich partner hasthe biggest and bestpotential?Strategic alliancesMunicipalities,Governmenta new product line?Determine the likelihoodof a patent infringementLegal departmentCollect existing patentinfringement cases.Law enforcementWhat crimes occurredin a particularneighborhood last week?When are certain typesof crime most likely tooccur?Shipping servicesLogisticsWhich packages are onwhich delivery truck?How can I optimallyassign packages todelivery trucks so asto minimize the timerequired to deliver all thepackages?Marketing agencySocial media marketingHow many connectionsdoes a person have insocial media?Who is the optimalconnector in a socialnetwork?Sports/gamblingOdds makersWho won the mostWorld Cups in the historyof the game?Who will win next year’sWorld Cup, SuperBowl orWorld Series?TransportationLogisticscancelled the most lastyear.What cancellation willhave the least impact ontravelers and our bottomline?Conclusionprocesses to leverage their data assets, the true potential of data is still untouched in many organizations.Advanced analytics, particularly predictive analytics, can help reveal the future and optimize operations.Page: 6

RapidMiner provides software, solutions,analytics, data mining, and text mining. Weautomatically and intelligently analyze data(both structured and unstructured) – includingmultimedia and text – on a large scale.a cutting-edge, open-source data miner. Hundreds of thousands of applications are already in use in morewww.rapidminer.comRapidMiner USARapidMiner, Inc. HeadquartersCambridge, MA 02138RapidMiner United KingdomRapidMiner Ltd.Quatro House, Frimley RoadCamberley GU16 7ERUnited KingdomRapidMiner GermanyRapidMiner GmbHStockumer Str. 47544227 DortmundGermany

Advanced Analytics is “the analysis of all kinds of data using sophisticated quantitative methods (for example, statistics, descriptive and predictive data mining, simulation and optimization) to produce insights that traditional