微軟汽車行業(yè)大數(shù)據(jù)平臺(tái)建設(shè)探討

1、MICROSOFT CONFIDENTIAL1汽車行業(yè)汽車行業(yè) - - 數(shù)據(jù)驅(qū)動(dòng)的時(shí)代數(shù)據(jù)驅(qū)動(dòng)的時(shí)代MICROSOFT CONFIDENTIAL2汽車行業(yè)面臨的挑戰(zhàn)The Connected Marketing, Sales & ServiceOthersThe Connected VehicleConnected Product InnovationConnected OperationsTransformation OpportunitiesCustomer ExamplesMICROSOFT CONFIDENTIAL3信息化的縱深發(fā)展，帶來(lái)汽車行業(yè)的大變革數(shù)字化轉(zhuǎn)型 60%的客戶

2、的交互都是通過(guò)數(shù)字渠道和經(jīng)銷商網(wǎng)絡(luò) 車企需要增強(qiáng)他們的價(jià)值網(wǎng)絡(luò)，通過(guò)數(shù)字渠道的方式來(lái)洞察客戶全生命周期的管理。車輛信息Hub的轉(zhuǎn)型 智能互聯(lián)車輛 到2020你那互聯(lián)車輛的滲透率將從20%上升到90% 開(kāi)放的生態(tài)環(huán)境將給車企帶來(lái)更多的挑戰(zhàn)，也給車企提供更多的服務(wù)機(jī)會(huì) 獨(dú)特的互聯(lián)客戶體驗(yàn)是未來(lái)的主要方向高度互聯(lián)的業(yè)務(wù)變化要求無(wú)處不在的客戶體驗(yàn)需要MICROSOFT CONFIDENTIAL4汽車行業(yè)信息化的趨勢(shì)為四個(gè)“Connected”今天的技術(shù)趨勢(shì)正在引爆機(jī)會(huì)！移動(dòng)企業(yè)社交大數(shù)據(jù)云計(jì)算物聯(lián)網(wǎng)Connected Marketing, Sales & Service客戶更愿意通過(guò)數(shù)字渠道結(jié)

3、合經(jīng)銷商網(wǎng)絡(luò)來(lái)做決定Connected Operations成本和風(fēng)險(xiǎn)管理依然是企業(yè)制造和運(yùn)營(yíng)的關(guān)鍵因素Connected Product Innovation智能互聯(lián)的產(chǎn)品驅(qū)動(dòng)了開(kāi)發(fā)過(guò)程中的創(chuàng)新Connected Vehicle驅(qū)動(dòng)更多的個(gè)性化客戶體驗(yàn)和持續(xù)服務(wù)MICROSOFT CONFIDENTIAL5531531 應(yīng)用46004600 接口 34003400 晚間批量任務(wù)沒(méi)有沒(méi)有 “端到端” 的流程的流程. 海量非結(jié)構(gòu)化數(shù)據(jù)支持效率: 一般一個(gè)一般一個(gè)員員工只能支持工只能支持1717個(gè)個(gè)應(yīng)應(yīng)用用車企的“Connected”首先意味著大量業(yè)務(wù)數(shù)據(jù)的整合MICROSOFT CONFIDEN

4、TIAL6所有的數(shù)據(jù)都是復(fù)雜、耗時(shí)和難以獲取數(shù)量 數(shù)據(jù)爆炸性增長(zhǎng)集成 各自為政的數(shù)據(jù)孤島質(zhì)量 數(shù)據(jù)質(zhì)量不可靠行動(dòng) 難以快速獲得數(shù)據(jù)價(jià)值然而：業(yè)務(wù)整合只是一個(gè)過(guò)程MICROSOFT CONFIDENTIAL7設(shè)計(jì)優(yōu)化供應(yīng)鏈合作新型服務(wù)能力優(yōu)化資源配置通過(guò)對(duì)客戶的洞察來(lái)更好更客戶互動(dòng)捕獲細(xì)致的數(shù)據(jù)和優(yōu)化性能優(yōu)化設(shè)備和流程性能2. 2. 復(fù)雜的數(shù)據(jù)流和監(jiān)控規(guī)則復(fù)雜的數(shù)據(jù)流和監(jiān)控規(guī)則7. 7.機(jī)器自學(xué)習(xí)機(jī)器自學(xué)習(xí)1. 1. 可擴(kuò)展的數(shù)據(jù)鏈接和數(shù)據(jù)獲取可擴(kuò)展的數(shù)據(jù)鏈接和數(shù)據(jù)獲取3. 3. 可按地域的云存儲(chǔ)可按地域的云存儲(chǔ)4. 4. 大數(shù)據(jù)大數(shù)據(jù)5. 5. BI BI商業(yè)智能商業(yè)智能5. 5. BI BI

5、商業(yè)智能商業(yè)智能5. 5. BI BI商業(yè)智能商業(yè)智能6. 6. 客戶和服務(wù)管理客戶和服務(wù)管理6. 6. 客戶和服務(wù)管理客戶和服務(wù)管理提供差異化的客戶體驗(yàn)車企需要構(gòu)建這樣的大數(shù)據(jù)平臺(tái)和業(yè)務(wù)互聯(lián)MICROSOFT CONFIDENTIAL8互聯(lián)大數(shù)據(jù)平臺(tái)帶來(lái)的價(jià)值客戶親密度情景體驗(yàn) | 跨設(shè)備體驗(yàn)關(guān)系生命周期管理crm | plm | scm | dms機(jī)器自學(xué)習(xí)預(yù)測(cè) | 分析云平臺(tái) + 基礎(chǔ)架構(gòu)全局化 | 可擴(kuò)展DataMICROSOFT CONFIDENTIAL9汽車行業(yè)互聯(lián)的大數(shù)據(jù)平臺(tái)方案The Connected Marketing, Sales & ServiceOthersT

6、he Connected VehicleConnected Product InnovationConnected OperationsTransformation OpportunitiesCustomer ExamplesMICROSOFT CONFIDENTIAL10Hadoop 集群采集和預(yù)處理車聯(lián)網(wǎng)Connected carAPPsLOBCRMSAPEDW 數(shù)據(jù)倉(cāng)庫(kù)Streaming實(shí)時(shí)分析互聯(lián)網(wǎng)Internet官方網(wǎng)站社區(qū)輿論情報(bào)車聯(lián)網(wǎng)RFID車載娛樂(lè)終端試驗(yàn)車日志處理企業(yè)大數(shù)據(jù)中心數(shù)據(jù)利用ETL分析、展現(xiàn)和其他應(yīng)用Connector算法和模型R Server /機(jī)器學(xué)習(xí) /分析挖

7、掘檢測(cè)監(jiān)控預(yù)測(cè)維護(hù)分析展現(xiàn)市場(chǎng)營(yíng)銷大數(shù)據(jù)中心平臺(tái)整體架構(gòu)MICROSOFT CONFIDENTIAL11Landing LayerBatch LayerServing LayerSpeed LayerRaw Data StorageStatic Views PrecomputingPrecomputingAd-hoc Batch ViewsStatic Batch Views實(shí)時(shí)數(shù)據(jù)流Raw Data Raw Data StorageStoragePre-computingPre-computingBatch ViewsBatch ViewsLegend:Legend:Layer bounda

8、ryData flow (with direction indicated)Data exchangeData LandingData Landing數(shù)據(jù)采集數(shù)據(jù)處理數(shù)據(jù)展現(xiàn)Indexing and ViewsReal-Time Processing, AggregationsStreamingStreaming實(shí)時(shí)數(shù)據(jù)流實(shí)時(shí)數(shù)據(jù)流Processing Microsoft Microsoft R ServerR ServerSQL Server BISQL Server BIBI to WebBI to WebSQL ModelingSQL ModelingData AreaData Area

9、大數(shù)據(jù)中心平臺(tái)邏輯架構(gòu)企業(yè)私有云平臺(tái)基礎(chǔ)架構(gòu)MICROSOFT CONFIDENTIAL12大數(shù)據(jù)中心三大組成部分基礎(chǔ)架構(gòu)基礎(chǔ)架構(gòu)- -私有云平臺(tái)私有云平臺(tái)數(shù)據(jù)存儲(chǔ)及運(yùn)算數(shù)據(jù)存儲(chǔ)及運(yùn)算-Hadoop-Hadoop平臺(tái)平臺(tái)數(shù)據(jù)分析及展示數(shù)據(jù)分析及展示-R-Server/BI-R-Server/BI數(shù)據(jù)分析平臺(tái)數(shù)據(jù)分析平臺(tái)1 12 23 3MICROSOFT CONFIDENTIAL13虛擬化管理工具服務(wù)器存儲(chǔ)網(wǎng)絡(luò)設(shè)備服務(wù)流程平臺(tái) System Center Service Manager數(shù)據(jù)中心管理員（資源管理）應(yīng)用管理員（應(yīng)用擴(kuò)容）最終用戶（服務(wù)申請(qǐng)）知識(shí)庫(kù)報(bào)表 系統(tǒng)管理數(shù)據(jù)、流程總線配置管理

10、工具監(jiān)控管理工具門(mén)戶管理平臺(tái) SharePoint Server災(zāi)備管理工具傳統(tǒng)物理資源虛擬化資源微軟私有云(IaaS)平臺(tái)邏輯架構(gòu)MICROSOFT CONFIDENTIAL14構(gòu)建統(tǒng)一管理的私有云平臺(tái)分配專用和共享的資源數(shù)據(jù)中心資源數(shù)據(jù)中心資源邏輯與標(biāo)準(zhǔn)化基礎(chǔ)架構(gòu)多樣化云抽象委派的容量標(biāo)準(zhǔn)化服務(wù)開(kāi)發(fā)生產(chǎn)MICROSOFT CONFIDENTIAL15私有云的運(yùn)營(yíng)數(shù)據(jù)中心管理者監(jiān)控運(yùn)維虛擬服務(wù)器私有云物理服務(wù)器網(wǎng)絡(luò)與網(wǎng)絡(luò)設(shè)備監(jiān)控排錯(cuò)監(jiān)視運(yùn)維數(shù)據(jù)保護(hù)補(bǔ)丁安裝MICROSOFT CONFIDENTIAL16基礎(chǔ)架構(gòu)監(jiān)控MICROSOFT CONFIDENTIAL17自動(dòng)化修復(fù)IR2667Orch

11、estratorSQL Server 1SQL Server 2Service managerOperations managerVirtual machine managerData protection managerMICROSOFT CONFIDENTIAL18動(dòng)態(tài)優(yōu)化資源優(yōu)化配置增強(qiáng)可用性降低能耗MICROSOFT CONFIDENTIAL19更新管理更合規(guī)，保持最新?tīng)顟B(tài)可配合現(xiàn)有更新服務(wù)器使用無(wú)需停機(jī)啟用功能管理基準(zhǔn)線掃描服務(wù)器修補(bǔ)服務(wù)器管理例外虛擬宿主機(jī)Virtual machine manager server更新服務(wù)器更新編錄更新MICROSOFT CONFIDENTIAL2

12、0服務(wù)交互自動(dòng)化Configuration Manager Configuration Manager 數(shù)據(jù)庫(kù)（數(shù)據(jù)庫(kù)（CMDBCMDB）運(yùn)行手冊(cè)模板服務(wù)虛擬機(jī)宿主機(jī)用戶促進(jìn)標(biāo)準(zhǔn)化與合規(guī)性模版Virtual machine manager 庫(kù)發(fā)布用戶角色數(shù)據(jù)中心管理者用戶MICROSOFT CONFIDENTIAL21自動(dòng)化體檢MICROSOFT CONFIDENTIAL22記賬/收費(fèi)MICROSOFT CONFIDENTIAL23此階段的存儲(chǔ)數(shù)據(jù)實(shí)時(shí)同步由硬件產(chǎn)商提供解決方案，最終實(shí)現(xiàn)同區(qū)域的數(shù)據(jù)中心機(jī)房間的冗余，兩邊都能實(shí)時(shí)寫(xiě)入相同的數(shù)據(jù)副本，以保證故障產(chǎn)生時(shí)能平滑地切換。本地災(zāi)備：主站點(diǎn)

13、數(shù)據(jù)中心機(jī)房的冗余MICROSOFT CONFIDENTIAL24可備份到公有云應(yīng)用創(chuàng)新應(yīng)用創(chuàng)新Microsoft AzurePublic, Global, Shared Datacenters私有云數(shù)據(jù)中心 Core Software InfrastructureIaaSPaaSComputeStorageNetworkAzure Management PortalAzure Management Portal Cloud-inspired infrastructurePowered by Windows Server, Hyper-V, System Center (future Azur

14、e Stack)Windows Azure PackWindows Azure Pack(future Azure Stack Portal)IaaSPaaSComputeStorageNetworkAzure Global DatacentersDatacenter InfrastructureLinuxWindows ServerLinuxWindows ServerSystem Center + OMS從私有云擴(kuò)展到混合云MICROSOFT CONFIDENTIAL26大數(shù)據(jù)中心三大組成部分基礎(chǔ)架構(gòu)基礎(chǔ)架構(gòu)- -私有云平臺(tái)私有云平臺(tái)數(shù)據(jù)存儲(chǔ)及運(yùn)算數(shù)據(jù)存儲(chǔ)及運(yùn)算-Hadoop-Hadoop

15、平臺(tái)平臺(tái)數(shù)據(jù)分析及展示數(shù)據(jù)分析及展示-R-Server/BI-R-Server/BI數(shù)據(jù)分析平臺(tái)數(shù)據(jù)分析平臺(tái)1 12 23 3MICROSOFT CONFIDENTIAL27為什么需要Hadoop數(shù)據(jù)來(lái)源數(shù)據(jù)系統(tǒng)數(shù)據(jù)訪問(wèn)業(yè)務(wù)分析定制應(yīng)用現(xiàn)有數(shù)據(jù)數(shù)據(jù)庫(kù)運(yùn)營(yíng)應(yīng)用新數(shù)據(jù)有限的數(shù)據(jù)不足以管理現(xiàn)有的數(shù)據(jù)，更不用說(shuō)新數(shù)據(jù)，同時(shí)還要維護(hù)規(guī)模性能。數(shù)據(jù)首先必須是結(jié)構(gòu)化的。有限的洞察有限的訪問(wèn)和較差的數(shù)據(jù)視角導(dǎo)致較差的分析和低采用度。 很難將洞察轉(zhuǎn)化為行動(dòng)。復(fù)雜的架構(gòu)更多的數(shù)據(jù)，更多的用戶，更多的工具制造了復(fù)雜性。統(tǒng)一安全中的漏洞讓不良行為者可以有訪問(wèn)權(quán)。傳統(tǒng)方式的設(shè)計(jì)是為結(jié)構(gòu)化：結(jié)構(gòu)化的數(shù)據(jù)結(jié)構(gòu)化的分析結(jié)構(gòu)化

16、的流程MICROSOFT CONFIDENTIAL28為什么需要Hadoop 數(shù)據(jù)類型的限制Enterprise Data Warehouse應(yīng)用數(shù)據(jù)源結(jié)構(gòu)化非結(jié)構(gòu)化加載Staging Environment 傳統(tǒng)架構(gòu)Enterprise Data Warehouse服務(wù)ELT歸檔BI 應(yīng)用建模報(bào)表ETLStorage #1Storage #2Storage NIngest處理加載1MICROSOFT CONFIDENTIAL29為什么需要Hadoop 數(shù)據(jù)類型的限制2) 處理能力的限制Enterprise Data Warehouse應(yīng)用數(shù)據(jù)源結(jié)構(gòu)化非結(jié)構(gòu)化IngestStaging Env

17、ironment 傳統(tǒng)架構(gòu)Enterprise Data WarehouseServeELTArchiveBI 系統(tǒng)建模報(bào)表ETLStorage #1Storage #2Storage NIngestProcessLoad122MICROSOFT CONFIDENTIAL30為什么需要Hadoop 數(shù)據(jù)類型的限制2) 處理能力的限制3) 數(shù)據(jù)容量的限制Enterprise Data WarehouseApplicationsData SourcesStructuredUnstructuredIngestStaging Environment 傳統(tǒng)架構(gòu)Enterprise Data Wareho

18、useServeELTArchiveBI SystemModelingReportingETLStorage #1Storage #2Storage NIngestProcessLoad1223MICROSOFT CONFIDENTIAL31新的方式1)接收更多的數(shù)據(jù)ApplicationsData SourcesStructuredUnstructuredStaging Environment Hadoop架構(gòu)Enterprise Data WarehouseEDHIngestIngestActive Structured DataServeServeELTArchive Load1ETLB

19、I SystemModelingReportingMICROSOFT CONFIDENTIAL32新的方式1接收更多的數(shù)據(jù)2)強(qiáng)大的并發(fā)處理能力ApplicationsData SourcesStructuredUnstructuredStaging Environment Hadoop架構(gòu)Enterprise Data WarehouseEDHIngestIngestActive Structured DataServeServeELTArchive Load21ETLBI SystemModelingReportingMICROSOFT CONFIDENTIAL33新的方式1) 接收更多的

20、數(shù)據(jù)2) 強(qiáng)大的并發(fā)處理能力3) 全量數(shù)據(jù)在線ApplicationsData SourcesStructuredUnstructuredStaging Environment Hadoop架構(gòu)Enterprise Data WarehouseEDHIngestIngestActive Structured DataServeServeELTArchive Load231ETLBI SystemModelingReportingMICROSOFT CONFIDENTIAL34Hadoop-企業(yè)級(jí)的數(shù)據(jù)HUBHadoop 提供：一個(gè)地方用于無(wú)限任何類型的數(shù)據(jù)統(tǒng)一的，多框架的數(shù)據(jù)訪問(wèn)領(lǐng)先的性能和可

21、擴(kuò)展性企業(yè)要求：高性能開(kāi)源，開(kāi)放標(biāo)準(zhǔn)完整的系統(tǒng)和數(shù)據(jù)管理企業(yè)級(jí)的安全性全面數(shù)據(jù)監(jiān)管安全和管理無(wú)限存儲(chǔ)處理發(fā)現(xiàn)制模服務(wù)部署靈活性現(xiàn)場(chǎng)設(shè)備過(guò)程化系統(tǒng)共有云私有云混合云MICROSOFT CONFIDENTIAL35Hadoop: 擴(kuò)展性 & 靈活性 存儲(chǔ) & 計(jì)算The Hadoop傳統(tǒng)數(shù)據(jù)倉(cāng)庫(kù)$30,000+ per TB昂貴而遙不可及 擴(kuò)展困難 網(wǎng)絡(luò)成為不可避免的瓶頸 只能處理結(jié)構(gòu)化關(guān)系型數(shù)據(jù) 很難增加新的字段和數(shù)據(jù)類型昂貴的、專用的、“可靠的”服務(wù)器昂貴的軟件許可證Network數(shù)據(jù)存儲(chǔ)(SAN, NAS)計(jì)算(RDBMS, EDW)$300-$1,000 per TB經(jīng)濟(jì)且

22、可以企及 可無(wú)限平行擴(kuò)展 網(wǎng)絡(luò)不再是瓶頸 輕松攝取任何類型的數(shù)據(jù) 靈活的讀取時(shí)檢查數(shù)據(jù)類型的訪問(wèn)方式商業(yè)化的“不怎么可靠”的服務(wù)器混合的開(kāi)源軟件計(jì)算(CPU)內(nèi)存存儲(chǔ)(Disk)zzMICROSOFT CONFIDENTIAL36大數(shù)據(jù)中心三大組成部分基礎(chǔ)架構(gòu)基礎(chǔ)架構(gòu)- -私有云平臺(tái)私有云平臺(tái)數(shù)據(jù)存儲(chǔ)及運(yùn)算數(shù)據(jù)存儲(chǔ)及運(yùn)算-Hadoop-Hadoop平臺(tái)平臺(tái)數(shù)據(jù)分析及展示數(shù)據(jù)分析及展示-R-Server/BI-R-Server/BI數(shù)據(jù)分析平臺(tái)數(shù)據(jù)分析平臺(tái)1 12 23 3MICROSOFT CONFIDENTIAL37微軟數(shù)據(jù)產(chǎn)品家族RDBMSFlat/Excel fileXMLApplica

23、tionStream DataDataWarehouseSSRSSharePointWeb BrowserDatazen ServiceDatazen AppSSRS forSharepointSocial DataSSASData Mining ResearchMulti-Dimension ModelAPS / PDWSSISSQL ServerTabular ModelExcelPowerPivot ModelPerformance PointAzureService BusEvent HubHDInsightAzure SQL DBMachine LearningStream Insi

24、ghtSource of Data / BrowserWeb Application / Excel / O365SQL Server Service ComponentImportant Function / Add-inService on Azure PlatformAPS / PDWStreamAnalysisRevolutionRDocument DBSupport to EmbeddedR ScriptPower BI SaaSSolutionPower BISaaSPower BI SitePowerPivot Add-inPower View Add-inPower Query

25、Power ViewPower MapAzure PortalData FactoryPower BI DesktopAzure SQL DWMICROSOFT CONFIDENTIAL38微軟提供從自建數(shù)據(jù)中心到云端的完整解決方案MICROSOFT CONFIDENTIAL39 Analyst ReportsPeriod Leadership PositionGartner Magic Quadrant For Data Warehouse Database Management System For AnalyticsFeb 2015LeaderGartner Magic Quadrant

26、 For Business Intelligence And Analytics PlatformsFeb 2015LeaderGartner Magic Quadrant For Operational Database Management SystemOct 2015Leader微軟今天在商務(wù)智能和數(shù)據(jù)倉(cāng)儲(chǔ)中的領(lǐng)導(dǎo)地位MICROSOFT CONFIDENTIAL40HDInsightHortonworks Data Platform(HDP)微軟和HadoopMICROSOFT CONFIDENTIAL41語(yǔ)言平臺(tái)#1 Procedural Language optimized for

27、Statistics and Data Science A Data Visualization Framework Provided as Open Source社區(qū) 3M+ Statistical Analysis and Machine Learning Users Taught in Most University Statistics Programs Active User Groups Across the World生態(tài)環(huán)境 CRAN: 7500+ Freely Available Algorithms, Test Data and Evaluations Many Appli

28、cable to Big Data If Scaled什么是 R?Tool Use for Data ScienceTool Use for Data ScienceOReilly Data Science Survey 2014 (max=80%)MICROSOFT CONFIDENTIAL42Strategic RationaleStrategic RationaleThe leading providerof advanced analytics advanced analytics software and servicessoftware and servicesbased on o

29、pen source R, since 2007 REVOLUTION RREVOLUTION R: The enterprise-grade predictive analytics application platform based on the R language微軟和Revolution AnalyticsMICROSOFT CONFIDENTIAL43基于硬盤(pán)的擴(kuò)展性 多線程商業(yè)支持5000個(gè)算法分析包+ RevoScaleR 大數(shù)據(jù)分析包商業(yè)許可內(nèi)存限制單線程開(kāi)源社區(qū)支持5000個(gè)算法分析包開(kāi)源算法執(zhí)行風(fēng)險(xiǎn)大數(shù)據(jù)大數(shù)據(jù)分析速度分析速度企業(yè)級(jí)企業(yè)級(jí)廣度和深度廣度和深度商業(yè)上可行性

30、商業(yè)上可行性43為什么要Revolution AnalyticsR是開(kāi)源的數(shù)據(jù)分析軟件，并驅(qū)動(dòng)分析創(chuàng)新，但是.對(duì)企業(yè)而言，R有諸多局限性MICROSOFT CONFIDENTIAL44Revolution R Enterprise 7.4 (RRE)全套大數(shù)據(jù)分析平臺(tái)全套大數(shù)據(jù)分析平臺(tái)R+CRANR+CRANRevolution R OpenRevolution R OpenDistributedRDistributedRDeployRDeployRDevelopRDevelopRScaleRScaleRConnectRConnectR高性能開(kāi)源R plusplus:可連接到大數(shù)據(jù)對(duì)象的數(shù)據(jù)源

31、大數(shù)據(jù)高級(jí)分析多平臺(tái)支持 開(kāi)發(fā)和生產(chǎn)In-Hadoop 和 in-Teradata 預(yù)測(cè)模型為數(shù)據(jù)分析和模型開(kāi)發(fā)人員設(shè)計(jì)的IDE安全的，可擴(kuò)展的R模型部署技術(shù)支持，培訓(xùn)和專業(yè)服務(wù)微軟整合Revolution Analytics為Microsoft R-ServerMICROSOFT CONFIDENTIAL45Microsoft R Server provides a unique opportunity to deliver our advanced analytics capabilities to customers who have already invested in storin

32、g their data on non Microsoft platforms like Hadoop, Teradata and LinuxMicrosoft R-Server將繼續(xù)對(duì)非微軟平臺(tái)的支持MICROSOFT CONFIDENTIAL46Best of Both Worlds靈活性&敏捷性O(shè)perationalize R scripts via stored procedures通過(guò)存儲(chǔ)過(guò)程編寫(xiě)R腳本R code portability across platforms提供R代碼的可移植性性能&可擴(kuò)展 Reduce data movement by bringing

33、 compute to the data通過(guò)對(duì)數(shù)據(jù)進(jìn)行計(jì)算來(lái)降低數(shù)據(jù)的移動(dòng) In-Memory Analytics內(nèi)存中分析 In-DB Parallelized R Analytics數(shù)據(jù)庫(kù)并行R分析成本效率 R Services built-in內(nèi)置R服務(wù) No proprietary hardware requirement沒(méi)有專有的硬件需求用用SQL 快速查詢快速查詢 & in-memory ColumnStore indexes用用 R于數(shù)據(jù)探索、于數(shù)據(jù)探索、 預(yù)測(cè)建模、預(yù)測(cè)建模、 Scoring和可視化和可視化使用SQL Server的R Service構(gòu)建企業(yè)智能應(yīng)用Mi

34、crosoft R-Server作為SQL Server2016內(nèi)置的服務(wù)MICROSOFT CONFIDENTIAL47Data import Delimited, Fixed, SAS, SPSS, OBDCVariable creation & transformationRecode variablesFactor variablesMissing value handlingSort, Merge, SplitAggregate by category (means, sums)Min / Max, Mean, Median (approx.)Quantiles (appro

35、x.)Standard DeviationVarianceCorrelationCovarianceSum of Squares (cross product matrix for set variables)Pairwise Cross tabsRisk Ratio & Odds RatioCross-Tabulation of Data (standard tables & long form)Marginal Summaries of Cross TabulationsChi Square TestKendall Rank CorrelationFishers Exact

36、 TestStudents t-TestSubsample (observations & variables)Random Sampling數(shù)據(jù)預(yù)處理統(tǒng)計(jì)檢驗(yàn)抽樣描述性統(tǒng)計(jì)Sum of Squares (cross product matrix for set variables)Multiple Linear RegressionGeneralized Linear Models (GLM) exponential family distributions: binomial, Gaussian, inverse Gaussian, Poisson, Tweedie. Standa

37、rd link functions: cauchit, identity, log, logit, probit. User defined distributions & link functions.Covariance & Correlation MatricesLogistic RegressionClassification & Regression TreesPredictions/scoring for modelsResiduals for all models預(yù)測(cè)模型K-MeansDecision TreesDecision ForestsGradie

38、nt Boosted Decision TreesNave Bayes聚類分析分類模擬變量選擇Stepwise RegressionSimulation (e.g. Monte Carlo)Parallel Random Number Generation 結(jié)合開(kāi)源R New in New in v7.3v7.3PEMA-R APIrxDataSteprxExec微軟R-Server提供大量的算法和函數(shù)MICROSOFT CONFIDENTIAL48Data Warehouse LayerRaw Data LayerSQL Integration ServicesData Analysis L

39、ayerSQL Analysis ServicesSQL Database ServicesData Visualization LayerPower BIn數(shù)據(jù)展現(xiàn)層，提供多樣化的報(bào)表展示應(yīng)用。相關(guān)組件：Reporting Service / Datazen / PowerBIn數(shù)據(jù)分析層，提供數(shù)據(jù)建模、多維分析，數(shù)據(jù)挖掘等多種功能，優(yōu)化查詢性能。相關(guān)組件：Analysis Servicesn數(shù)據(jù)倉(cāng)庫(kù)層，從各個(gè)數(shù)據(jù)源集成數(shù)據(jù)，清洗數(shù)據(jù)，并存儲(chǔ)數(shù)據(jù)做進(jìn)一步處理相關(guān)組件：Database Engine，Integration ServicesMobile SolutionReporting Se

40、rvice微軟提供多樣化的BI展示方式MICROSOFT CONFIDENTIAL49微軟不斷增強(qiáng)Mobile BI的功能Reporting ServicesSharepointOnlineAnalysis ServicesData WarehouseStream InsightData WarehouseCrawlerETLSensorsDevicePower BIWeb ReportingApp ToolIaaSPaaSSaaSApp on AzureMICROSOFT CONFIDENTIAL50Mobile BIMICROSOFT CONFIDENTIAL51微軟PowerBIMICR

41、OSOFT CONFIDENTIAL52客戶案例The Connected Marketing, Sales & ServiceOthersThe Connected VehicleConnected Product InnovationConnected OperationsTransformation OpportunitiesCustomer ExamplesMICROSOFT CONFIDENTIAL53微軟在汽車行業(yè)的客戶和合作伙伴Delivering core technology platforms and solutions for over thirty yearsP

42、roduct Development and InnovationGlobalOperationsTelematics 2.0 and the Connected Car Marketing, Sales and Service MICROSOFT CONFIDENTIAL54GPSGPSTCU遠(yuǎn)程控制單元遠(yuǎn)程控制單元汽車總線大數(shù)據(jù)集群大數(shù)據(jù)集群高速采集服務(wù)器高速采集服務(wù)器車輛配置信息實(shí)時(shí)總線數(shù)據(jù)GPS數(shù)據(jù)3G3G移動(dòng)網(wǎng)絡(luò)移動(dòng)網(wǎng)絡(luò)1.車輛遠(yuǎn)程跟蹤診斷系統(tǒng)實(shí)時(shí)監(jiān)控實(shí)時(shí)監(jiān)控遠(yuǎn)程指令遠(yuǎn)程指令報(bào)表分析報(bào)表分析大數(shù)據(jù)在汽車行業(yè)的應(yīng)用案例MICROSOFT CONFIDENTIAL55需求 新車型量產(chǎn)前，采集并保存路試車輛的行駛數(shù)據(jù)。 每輛路試車輛每天產(chǎn)生的數(shù)據(jù)從50M到1G不等，取決于采集的數(shù)據(jù)類型和