Presentation is loading. Please wait.

Presentation is loading. Please wait.

第六章 資料倉儲與採礦技術 6.1 資料倉儲與採礦定義 6.2 資料採礦之步驟與技術分類 6.3 資料採礦在顧客關係管理之應用

Similar presentations


Presentation on theme: "第六章 資料倉儲與採礦技術 6.1 資料倉儲與採礦定義 6.2 資料採礦之步驟與技術分類 6.3 資料採礦在顧客關係管理之應用"— Presentation transcript:

1 第六章 資料倉儲與採礦技術 6.1 資料倉儲與採礦定義 6.2 資料採礦之步驟與技術分類 6.3 資料採礦在顧客關係管理之應用
第六章 資料倉儲與採礦技術 6.1 資料倉儲與採礦定義 6.2 資料採礦之步驟與技術分類 6.3 資料採礦在顧客關係管理之應用 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

2 Chapter 6 Date Warehousing & Data Mining
6.1 Definitions of data warehousing & data mining 6.2 Approach and technologies of data mining 6.3 Application of data mining in CRM 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

3 6.1 資料倉儲與採礦定義 商業智慧之背景 E化時代之資訊特點: E化時代下,電子資料量每年成長1~3倍 僅7%之資料真正被分析、運用者
資訊應用需求:轉化龐大無序之資料為資訊、知識,進而供有智慧、經驗者運用 具體目標: 協助組織擷取/歸納/解釋/分析資料 從事獲利分析/關係行銷/顧客管理 領域應用:客戶貢獻度分析、市場區隔、信用風險管理、交叉銷售分析、產品/投資組合獲利分析 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

4 6.1 Definitions of Data Warehousing & Data Mining
Background of Business Intelligence (BI) Information features in the e-century: In E era, the amount of electrical data grows 1 to 3 times per year. only 7% of existing data are properly analyzed and applied. Demand for information application: data -> information -> knowledge -> wisdom. Goals: To assist organization in obtaining / generalizing / explaining /analyzing data. Profit analysis / Relationship marketing / Customer management Application domain: customer contribution analysis / market segmentation / risk management / cross-sells analysis / product profile analysis / investment portfolio analysis 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

5 6.1 資料倉儲與採礦定義 商業智慧(BI)定義 定義:對企業營運資訊迅速解讀與推理之能力,以提升企業決策品質、改善營運績效。 特點:
依公司既有資訊此分析業務發展趨勢 以公司既有資訊進行決策支援 參與人員可及時獲取其職責所需之知識 Provide the Right Information to the Right Persons at the Right Time. BI技術可協助企業統計、挖掘、分析與轉化隱含於大量數據資料背後之知識,以支援企業相關決策 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

6 6.1 Definitions of Data Warehousing & Data Mining
Definition of Business Intelligence (BI) Definition: The ability to rapidly analyze and synthesize enterprise data in order to improve its decision quality and business performance. Features: Analyze business development trend. Decision supports. Provide the Right Information to the Right Persons at the Right Time. BI techniques can provide statistical analysis, data mining and analysis, and transform great quantity of data into meaningful knowledge to support the decision making. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

7 6.1 資料倉儲與採礦定義 商業智慧之形成(1) BI的形成為Bottom-Up過程
公司組織架構中,資訊Bottom-Up方式進行收集,逐筆之記錄資料,形成原始資料(Row Data)。 基於決策需要,將資料萃取、解讀、處理、分析後,成為有組織、有價值之資訊(Organized Information)。 結合公司內部領域智能(domain know-how),將資訊轉化為知識(Knowledge)。 結合決策者的經驗與能力,將知識靈活應用,形成智慧(Wisdom)。 BI的形成為Bottom-Up過程 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

8 6.1 Definitions of Data Warehousing & Data Mining
Evolution of BI (1) In the organization, using the Bottom-Up process to collect and save the record that forms the Raw Data. On the basis of making a decision, selecting, reading, handling, analyzing the raw data the that obtain the Organized Information. Merging with organized information and domain know-how of the company that transforms the Knowledge. Added the experience of the decision-makers, fully utilized the knowledge that produce Wisdom. The approach to constitute BI is Bottom-Up process 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

9 6.1 資料倉儲與採礦定義 商業智慧之形成(2) Data—粗糙的事實(如實驗數據)
Information —融合脈絡之Data(如實驗結果) Knowledge — Information融合經驗(如實驗推論) Wisdom — 具啟發性之知識(如實驗結果應用) 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

10 6.1 Definitions of Data Warehousing & Data Mining
Evolution of BI (2) Data—raw data (ex: experiment data) Information —collection and arrangement of data (ex: experiment result) Knowledge — Integration of information and experience (ex: experiment conclusion) Wisdom — Heuristic knowledge (ex: application of experiment conclusion) 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

11 6.1 資料倉儲與採礦定義 商業智慧之應用 BI的應用是Top-Down過程 高層人員針對資訊進行決策之比例與重要性高。
越上層決策者故對於資訊/知識深度與廣度要求較高,更需較高之智慧運用之。 決策者運用智慧、經驗將決定策略後,下達予下部單位執行、實現。 決策者決策效率與資訊擷取、解讀(Bottom-Up)時效高度相關。 E化技術之應用可提升(1)資訊擷取/解讀與(2)策略制訂/傳遞/執行等效率。 BI的應用是Top-Down過程 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

12 6.1 Definitions of Data Warehousing & Data Mining
Application of BI (1) The high-level executives often need deeper knowledge to make crucial decisions. The application of e-technology can be used to improve (1) information extraction/understanding and (2)decision making /transaction. The applications of BI are Top-Down processes 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

13 6.1 資料倉儲與採礦定義 商業智慧主要應用主題 資料分析:發現大量資料之模式與趨勢(DW、DM)
智慧部署:運用IT技術使成員可進行商業分析與知識存取(KM) 地理空間分析:整合商業資料與地理/人口資訊(GIS) 資料視覺化:以圖形方式檢視資料,協助訂定更佳商業決策(KM、VR) 平衡計分卡分析:正確地檢視企業經營績效 專案管理分析:充分掌握資訊以訂定有關資源分配及專案選擇之決策(ERP、PM) 協同體系之智慧累積:運用您員工群體的見識與經驗(KM、PM) 業務/行銷分析:充分掌握銷售資料與趨勢(CRM) 客戶資訊分析:深入了解客戶行為與偏好(CRM) 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

14 6.1 Definitions of Data Warehousing & Data Mining
Application of BI (2) Data analysis: Discover the trend and model of the great quantity data(DW、DM) Intelligence disposition: Apply IT solution for commerce analysis and knowledge extraction.(KM) Geographical (spatial) analysis: Integrate with business data and geographic or population information. (GIS) Data Visualization: Use of the GUI to view data and to support the optimal business decisions. (KM、VR) 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

15 6.1 Definitions of Data Warehousing & Data Mining
Application of BI (3) BSC (Balance Score Card): Inspect the performance of the managing of enterprise correctly. Project management: Fully grasp information in order to make decision of resource allocation and project choosing (ERP、PM). Collaborative intelligence accumulation: Use the staff‘s experience (KM、PM). Business / marketing analysis:Fully control the sales data and trends (CRM). Customer information analysis:Understand customer‘s behavior and preferences (CRM). 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

16 6.1 資料倉儲與採礦定義 CRM—顧客關係管理定義 定義1:企業透過有效溝通,以瞭解、影響顧客行為,達成維繫與拓展客源之目標: 增加新顧客
防止既有顧客流失 提高顧客忠誠度 提高顧客獲利 定義2:是一種反覆將顧客資訊轉換為正面顧客關係的過程,藉由資訊科技,將資料轉化和視覺化,加速管理決策的實用性和速度。 結論:顧客關係管理乃能提供顧客接觸人員、資訊人員、行銷和銷售人員、及管理階層更多與顧客及銷售相關的企業智慧。 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

17 6.1 Definitions of Data Warehousing & Data Mining
Definition of CRM Definition 1: Through effective communication and understanding customer's behavior, enterprises can reach the connection and expand the customer source goal. Increase new customer. Take precautions against the existing customer lost. Improve customer's loyalty and the profit of the company. Definition 2:It is a kind of repeated course that changes customer‘s information into customer’s positive relation. Using of information technology, strengthen the practicability and speed of administrative decision. Conclusion:CRM can offer more relevant information to CSR, MIS, sales, and executive about relation BI. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

18 6.1 資料倉儲與採礦定義 CRM重點課題 因此應結合BI之理念與技術,有效收集、歸納、解析客服相關資訊,協助組織進行行銷相關之決策
為了與顧客關係真實存在,溝通必須為雙向的、整合的、有記錄的、有管理的。 若沒有顧客歷史資料、詳細交易記錄、有焦點和分類之溝通,就無法有效維持和顧客之關係。 Note:沒有絕對無用之資料 客服中心是顧客關係管理的第一步。 因此應結合BI之理念與技術,有效收集、歸納、解析客服相關資訊,協助組織進行行銷相關之決策 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

19 6.1 Definitions of Data Warehousing & Data Mining
Topics of CRM CRM should integrate all of the operations, members , transactions of the company. Customer relationship of CRM is not only trade (products or the service), but expect customers purchase continually or other valued behaviors. Communication must be two-way, integration, have records, management that make customer relationship is really exist. Contact Center is the first step of CRM. Integrate the concepts and technologies of BI; collection, extraction, analyze services information effectively, support decision making of the organization. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

20 6.1 資料倉儲與採礦定義 BI與CRM關聯 (1) 數位資訊包含企業多年累積之各種形式資料,內部/外部資訊,及各式與客戶或其他企業互動的累積經歷 這些累積資料、資訊與經歷極為寶貴,但卻十分繁雜 多數E化軟體負責處理交易(Transaction)相關數據資料,缺乏預測顧客行為模式及其對企業營運之影響。 導入BI之資料採礦理念,使企業檢視自身經營模式,並瞭解顧客特性 故CRM乃整合各行業之行銷與客戶管理之資訊系統及商業智慧(BI)決策分析機制。 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

21 6.1 Definitions of Data Warehousing & Data Mining
Relationship between BI and CRM (1) Digital information includes various kinds of data that enterprises have accumulated for many years. Internal and external information and various types of accumulated experiences with customers or other enterprises Most of enterprise information systems only deal with transactional data. BI data mining helps better understand enterprise operational models and customers characteristics. CRM integrates marketing and customer information systems and BI decision analysis systems. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

22 6.1 資料倉儲與採礦定義 商業智慧與CRM關聯(2)
Source: Source: 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

23 6.1 Definitions of Data Warehousing & Data Mining
Relationship between BI and CRM (2) Source: Source: 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

24 Customer & Sales Information Collection
商業智慧與CRM關聯(3) Customer & Sales Information Collection Contact Center Data Operation Information Collection ERP/Transaction Data Mining Information BI Operation Strategies Knowledge SFC/SCM Action CRM Action 運用BI觀念可將與客戶相關之銷售、服務資訊由Bottom-Up方式收集,再轉化為Top-Down之策略與應用 Wisdom 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

25 Customer & Sales Information Collection
Relationship between BI and CRM (3) Customer & Sales Information Collection Contact Center Data Operation Information Collection ERP/Transaction Data Mining Information BI Operation Strategies Knowledge SFC/SCM Action CRM Action Button-Up: collect related marketing, services and customers data Top-Down: strategies and applications. Wisdom 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

26 6.1 資料倉儲與採礦定義 Database Design Data Warehousing Data Mining
客服中心之資料建置與分析 Database Design Data Warehousing Data Mining Knowledge Recovery 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

27 6.1 Definitions of Data Warehousing & Data Mining
Steps of building and analyzing data of contact center Database Design Data Warehousing Data Mining Knowledge Recovery 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

28 6.1 資料倉儲與採礦定義 資料庫系統與其運作 「資料庫」乃將各種資料經過蒐集、整理之手法後,以適當方式加以組織化,再以有效的方式加以儲存。
利於資料後續搜尋、編輯、管理及再利用,以使資料使能提供多方面應用。 現今圖書館館藏訊息之蒐集、整理、組織化、儲存、搜尋及管理等,即為資料庫技術之典型利用。 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

29 6.1 Definitions of Data Warehousing & Data Mining
Database system application “Database” is an IT application that collects and marshals various data, and then saving them by effective and organizational approach. Database is widely used in various areas, i.e., library management. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

30 6.1 資料倉儲與採礦定義 資料庫系統相關詞彙 File(檔案):通常一資料檔案之建立乃針對特定應用及其對應所需之報表而來。隨著時日增加,可能會有許多檔案存在。 Record(記錄) Field(欄位) Character(文字) Entity(實體) Attribute(屬性;個別欄位名稱) 於一資料庫系統中,一Database含多筆Record、各筆Record以Field為單元,且各Field可為任意之資料型態,並以Data Dictionary進行內容分類。 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

31 6.1 Definitions of Data Warehousing & Data Mining
Database system (DBMS) related vocabularies File Record Field Character Entity Attribute A database system includes various Records that are form by Field. Field can be any data type, and using Data Dictionary to class the context. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

32 6.1 資料倉儲與採礦定義 檔案系統儲存資料之瓶頸 缺乏資料之安全性 缺乏資料之共用性 缺乏資料應用彈性 資料重複記錄
資料與程式具高度相關性,不易維護 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

33 6.1 Definitions of Data Warehousing & Data Mining
Bottleneck of file systems for data storage Lack security of the data Lack commonality of the data Lack flexibility of the data Repeat records Hard to maintain the system because of tight linkages between data and programs. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

34 6.1 資料倉儲與採礦定義 資料庫系統儲存資料之優點
故以資料庫系統取代檔案系統之資料儲存方式,增強資料儲存、索引搜尋、系統安全性與資料同步存取等特性。 其能針對每一不同的商業應用擷取資料,並產生其所需之報表。 具體優點乃可以減少檢索資料的時間與人力,又透過網路可更迅速地進行資料更新及管理。 具彈性化檢索功能,對大量資料的處理、搜尋所需資料的正確性與效率都會因此而大幅提昇。 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

35 6.1 Definitions of Data Warehousing & Data Mining
Advantages of database system for data storage Storage, search, indexing capabilities Extraction of information for different commerce applications. Generate the necessary reports. Internet access. Flexible search functions. Efficiently manage large amount of data. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

36 6.1 資料倉儲與採礦定義 資料庫運算(1) Selection乃自資料表中選出符合設定條件之行(Row)。
Projection乃自資料表中移除特定之列(Column)。 Product則是將兩Table相乘,若Table 1為N Rows、I Columns;Table 2為M Rows、J Columns,則其相乘結果產生一個具有N*M Rows、I+J Columns的新Table。 若兩Table分別為N Rows、M Rows,則Union結果形成N+M Rows(然除非兩Table的Schema相符,否則無意義)。 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

37 6.1 資料倉儲與採礦定義 資料庫運算(2) Set Difference乃找出於一Relation中存在,卻不在另一Relation中存在,兩Relation間的交集可以A-(A-B)獲得。 Join操作即是將兩個具有共同Field關聯之Tuple結合,即當Tuple中之欄位值與另一Tuple中之欄位值相等時則結合,可以Selection與Product定義之,如此即可以簡易資料表產生複雜之資料集合;相當於許多資料庫中之View功能,可將兩Row之資訊合併。 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

38 6.1 Definitions of Data Warehousing & Data Mining
Operations of database Selection is choosing the pass muster “Row” from table. Projection is moving out specific “Column” from table. Product is the multiplication of two tables. If Table 1 has N rows, I columns, and Table 2 has M rows, J columns, then the product is a new table that has N*M rows, I+J columns. If two table is N rows, M rows respectively, then Union result is N + M Rows . “Set Difference” is to find that data exists in one Relation and not exists in another. The interaction of two Relations is “A-(A-B)”. Join is to combine with two mutual “Fields”. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

39 6.1 資料倉儲與採礦定義 SQL (Structured Query Language)
SQL查詢語言乃為依據Relational Algebra衍生出之查詢語法。 其發展利於高階語言應用於資料庫資訊之擷取。 由IBM於1970年發展,且現已為ANSI標準。 除查詢RDBMS資料外,SQL亦提供定義Table Schema、資料建立與資料處理之功能。 如Create Table可指定屬性與資料欄位、Delete可刪除Row、Insert Into乃增加Row至一Table中)。 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

40 6.1 Definitions of Data Warehousing & Data Mining
SQL (Structured Query Language) SQL standards for Structured Query Language and is developed according to Relational Algebra. SQL is developed for allowing high level language to access the database. SQL is developed by IBM in 1970 and is an ANSI standard language. In addition to query the data of RDBMS, SQL can define the meaning of Table Schema, data built up and translation. Ex: “Creating Table” can set up data scheme and attribute, “Delete” can delete a row, “Insert Into” is added a row to a table. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

41 6.1 資料倉儲與採礦定義 RDBMS vs OODBMS
除RDBMS外,近期已逐漸發展物件導向資料庫管理系統(Object-Oriented Database Management System;OODBMS)。 在OODBMS環境下,所有物件之資料儲存於一處,並以唯一之Object ID作索引。 物件間之連結可直接由Object ID與Object Link追蹤(非以Join方式)。 OODBMS提供執行時之Schema Querying(Run-Time Schema Querying),即應用程式乃查詢物件的Meta-Data(元)資料。 現今應用Hybrid OO-Relational資料庫系統較多,其乃提供RDBMS系統中有物件之表示法,純OODBMS較為少用。 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

42 6.1 Definitions of Data Warehousing & Data Mining
RDBMS vs OODBMS Except RDBMS, DBM gradually develops OODBMS(Object-Oriented Database Management System) way. In the OODBMS environment, all objects save in one space and tracking by Object ID and Object Link. OODBMS provides really time Schema Querying(Run-Time Schema Querying) to query “Meta-Data” of application object. Using Hybrid OO-Relational database systems now is more, it offers the expression method with objects of RDBMS, pure OODBMS is comparatively used few. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

43 6.2 資料採礦之步驟與技術分類 客服中心資料庫設計 第一階段—資料需求規格釐清:不同使用者資訊需求確認
第二階段—概念設計:即建立「屬性-關係」模型,即Entity-Relation Model(簡稱ER Model) 第三階段—邏輯設計:乃將觀念架構轉換成所選定之資料庫管理系統,如關聯式資料庫管理系統之邏輯資料模式。 第四階段—實體設計:將邏輯資料模式轉換成特定硬體及DBMS適用之形式,通常乃決定資料儲存結構及檢索路徑。 資料庫正規化(Normalization):即自複雜、大量之資料中尋求其關聯性,而建立結構化表格。 主鍵(Primary Key)決定:資料庫中之鍵(Key),乃由一個主鍵(Primary Key)獨特地表示記錄的若干特性 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

44 6.2 Approach and technologies of data mining
Database design of the contact center First Step — The demand specification of DB should be clear out. Second Step — concept design:build up Entity-Relation Model (ER-Model) Third Step — logic desgn:transform into decided DBMS. (Ex: the logic data model of RDBMS) Fourth Step — reality design: transform the logic data model into hardware type of DBMS to decide data storage structure and search route. Normalization:seek relation, and set up structural format. Decide Primary Key:”Primary Key” represents the characteristic of this table uniquly. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

45 6.2 資料採礦之步驟與技術分類 ER Model範例 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

46 6.2 Approach and technologies of data mining
ER Model example 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

47 6.2 資料採礦之步驟與技術分類 客服中心之資料倉儲(1)
資料倉儲:一特大型資料庫,將來自不同作業系統之資料建立關聯性後,以具結構性方式加以儲存 自客服人員線上處理系統所得之資料,透過擷取、轉化與載入(Extraction、Transformation、Loading, ETL)等資訊技術,將資料置於資料倉庫中 必須確保倉庫中之資料之: 正確性(無錯誤資料參雜其中) 完整性(必要之資料皆被儲存) 相互整合 以交易主體(如顧客、產品)作為其儲存分類依據。 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

48 6.2 Approach and technologies of data mining
Data Warehousing of a contact center (1) Data Warehousing: is to bring together information from multiple operation systems as to provide a consistent database source for decision support. Collection of data by CSR (Customer Service representative) that is deal with extraction, transformation, loading, ETL and saving in the data base. Data is the database should be: correctness completeness Integration Classification according to translation (Ex: customers, products). 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

49 6.2 資料採礦之步驟與技術分類 客服中心之資料倉儲(2) 資料倉儲為決策支援系統之基礎,其技術具多元性與複雜性:
多維(Multi-Dimension)資料庫管理系統 主從架構(Client/Server Architecture) 中介軟體(Middleware) 圖形使用者介面(GUI) 資料複製(Replication) 平行處理(Parallel Processing) 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

50 6.2 Approach and technologies of data mining
Data Warehousing of a contact center (2) Data Warehouse is the base of Decision Support System (DSS) and the technologies are multi-dimension and complicated: Multi-Dimension Database Management System (DBMS) Client / Server Architecture Middleware softwre GUI (Graphical user interface) Replication Parallel Processing 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

51 6.2 資料採礦之步驟與技術分類 客服中心之資料倉儲(3) 基本特質: 以交易主體為導向 資料經過整合 各類使用者無法擅自更改資料
資料不斷隨時間而變化 收集之資料主旨在支援企業決策之制定 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

52 6.2 Approach and technologies of data mining
Data Warehousing of a contact center (3) Fundamental attributes: Classification according to translation Integration of multiple data Users can not change data without authorization Data changes with time constantly To help enterprise makes faster and better decisions. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

53 6.2 資料採礦之步驟與技術分類 客服中心之資料倉儲(4) 資料倉儲建置專案是否成功發揮其應有效益,與以下要素有高度相關性:
常設專案支援部門,支援專案之專業執行 界定明確目標及需求範圍,使資料倉儲平台規劃符合實需 高層支持與良好跨部門溝通管道,建立全公司共識 建構開放性之倉儲平台,並具備充分延展性及擴充性的倉儲架構 倉儲與前端系統之介面規格一致 前端系統資料品質高及具穩定性 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

54 6.2 Approach and technologies of data mining
Data Warehousing of a contact center (4) Whether construction the data warehousing project is succeeded and obtained its benefit, there are some essential factors that should have high dependence as following: Set up the support department and support the projects to carry out. Define the clear goal and demand range and let the data warehousing platform accord with the requirement. It should have the high-level executive support, a good trans-departmental communicative channel, and setting up the corporate accountability. The data warehousing platform must possesses opening, expanding, stable, attributes, keeping the same interface specification of the front system . 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

55 6.2 資料採礦之步驟與技術分類 資料倉儲導入程序 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

56 6.2 Approach and technologies of data mining
Data Warehousing Process 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

57 6.2 資料採礦之步驟與技術分類 資料倉儲具體效益 提昇使用者對應用系統之運用能力和資料分析的效率。
培養企業迅速取得資訊之能力以縮短管理階層理解事件發生之反應時間。 強化企業資訊集中與整合能力。 提供企業資料分析之新機制來支援Data Mining、OLAP等資料分析任務。 提升企業進行資訊趨勢分析能力。 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

58 6.2 Approach and technologies of data mining
Benefits of data warehousing Promote the operation ability and analysis efficiency of the system for users. Train enterprises the ability of obtaining information rapidly, shorten the reflect time of executive . Strengthen enterprise's information centralization and integration ability. Offer a new approach of analysis to enterprises, support analysis tasks , such as Data Mining , OLAP ,etc.. Improve enterprises to analyze the business trend. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

59 6.2 資料採礦之步驟與技術分類 資料倉儲趨勢 過去資訊技術的瓶頸,如跨平台資料難以整合、龐大資料量儲存及分析效率低等問題,皆可藉由資料倉儲技術逐漸克服。 過去偏重資料收集(資料庫技術應用),到現今強調價值資訊分析(資料採礦技術應用。 目前資料倉儲應用於企業客戶關係管理之課題將漸朝深度發展, 期能透過資訊採礦技術將大量客戶資料和交易數據轉化為有用資訊,讓決策主管和行銷人員可以隨時掌握客戶行為趨勢與變化,並針對個別狀況制定因應之行銷策略。 整體而言,透過資料倉儲技術運用,可將企業大量、無章之資料轉成商業知識,進而協助企業分析客戶習性及協助產品規劃,增加企業之競爭力與獲利機會。 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

60 6.2 Approach and technologies of data mining
Future trend of data warehousing The data warehousing technology overcomes the bottleneck of the information technology gradually, i.e., portability of cross-platform, analysis efficiency, and the huge amount data storage. The subject applied to CRM at present will be developed towards the depth value of information analysis gradually. Using the data warehousing technology, that can transform a great quantity, unorderly data into BI, and then help enterprises to analysis the customers and assistance in the product planning, increasing the competitiveness of enterprises and profit-making chance. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

61 6.2 資料採礦之步驟與技術分類 資料採礦 資料採礦(Data Mining)主要在探討如何於大量資料中發掘潛藏的有用資訊與趨勢,以提供決策人員進行決策參考。 資料採礦過程可視為資料庫管理之知識發掘。 知識發掘之過程包括資料選取、前置處理、資料轉換、資料採礦、解釋與評估。 與資料採礦相關之研究領域包括前述之資料庫技術、機器學習、人工智慧、專家系統、特徵識別、統計學及資料視覺化等課題 而目前較普遍運用之模式與技術包括決策樹、類神經網路、歸納式邏輯、貝氏網路(Baysian Network)、Nearest Neighbor、Attributed-Oriented Induction、Binary/Quantitative Association Rules等。 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

62 6.2 Approach and technologies of data mining
Data Mining discusses how to explore hiding useful information and trend among a large amount of data, in order to offer decision supports. Data Mining process can be viewed as a KDD process, including data selecting, pre-processing, data translation, data mining, explanation, and estimation. Related research areas with the data mining include DB technology, AI, expert system, data visualization, statistics. The popular models and technologies at present include decision tree, neural network, Baysian Network, Nearest Neighbor, Attributed-Oriented Induction, Binary/Quantitative Association Rules. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

63 6.3 資料採礦在顧客關係管理之應用 客服中心之資料採礦(1) 分析方向:
分類功能:乃按照分析對象之屬性,定義其分門類別,並建立類組(Class) 估計功能:乃根據既有連續性數值之相關屬性資料,獲致某一屬性未知之值 預測功能,乃根據對象屬性之過去觀察值推論該屬性之未來值,例如由顧客過去之閱讀趨勢預測其未來之消費項目 關聯分組功能:乃從所有物件決定那些相關物件應該列為同一群 分群功能:將異質母體中區隔為較具同質性之群組,相當於行銷所謂之區隔化(Segmentation) 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

64 6.3 Application of data mining to CRM
Data Mining of the contact center (1) Data Mining Functionalities: Classification: Classification is subdivided by assigning each element or record to a predefined class on the basis of a model developed through training on pre-classified examples. Estimation: Estimation deals with continuously valued outcomes and given some related input data to come up with a value for unknown continuous variable. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

65 6.3 Application of data mining to CRM
Data Mining of the contact center(2) Data Mining Functionalities: Prediction: Prediction is similar classification and estimation except that the records are classified according to some predicted future behavior or estimated future value according to observation value of the past. Affiliation Grouping: Affinity grouping is to determine which things go together. Clustering: Clustering is to segment a heterogeneous population into a number of more homogeneous subgroups or clusters. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

66 6.3 資料採礦在顧客關係管理之應用 資料採礦步驟 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

67 6.3 Application of data mining to CRM
Data Mining Approach 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

68 6.3 資料採礦在顧客關係管理之應用 資料採礦計畫 問題與目標確認:確認資料採礦所欲處理之潛在問題。
資料來源決定:決定資料採礦所需之基礎資料從何而來。 資料需求界定:即尋求與問題原因相關之資訊,並可能決定訪談之對象。 模型建立:可為簡單之OLAP或甚至複雜之類神經網路。 資料整理:基於不同資料模型有相亦之資料需求,故其資料整理方式亦當不同,並進行資料轉化。 利用資料倉儲支援:在資料採礦應用過程必須以資料倉儲作為支援。 軟體應用:利用既有交易資料及額外蒐集資料,透過專業之建構與分析軟體建立資料模型,含資料庫處理軟體或統計分析軟體。 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

69 6.3 Application of data mining to CRM
Data Mining Plan Confirm problem:Confirm the potential problems that are wanted to deal with. Decide data source:Determine where the basic data comes from . Decide requirement:Decide interview and seek the relevant problem data. Establish Model:OLAP or neurual network. Data arrangement:Transform the data because of different data models and data demand. Apply data warehousing:Data mining process must be supported with data warehousing. Software application:Set up model through the specialized statistical analysis software. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

70 6.3 資料採礦在顧客關係管理之應用 資料採礦模式 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

71 6.3 Application of data mining to CRM
Data Mining Model 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

72 6.3 資料採礦在顧客關係管理之應用 客服中心之資料倉儲(1)
資料倉儲:一特大型資料庫,將來自不同作業系統之資料建立關聯性後,以具結構性方式加以儲存 自客服人員線上處理系統所得之資料,透過擷取、轉化與載入(Extraction、Transformation、Loading, ETL)等資訊技術,將資料置於資料倉庫中 必須確保倉庫中之資料之: 正確性(無錯誤資料摻雜其中) 完整性(必要之資料皆被儲存) 相互整合 以交易主體(如顧客、產品)作為其儲存分類依據。 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

73 6.3 Application of data mining to CRM
Data Warehousing of the contact center(1) Data mining: A huge database using structural ways to store data that was been related from different operation systems. Extracting, transforming, loading, the data received from customer service personnel‘s on-line process system. Then store it in the database. Must ensure the data in the warehouse is: Correctness(There are not wrong materials mixed among them ) Completeness(essential data are all stored ) Combine each other Regard trade subject (such as customers, products) as its store categorized basis 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

74 6.3 資料採礦在顧客關係管理之應用 客服中心之資料採礦(2) 所收集之所有資訊都是寶貴資產,資料倉儲之重要性甚高。
資料採礦乃使用自動化或半自動化方式針對大量資料進行趨勢分析,並尋求對企業有意義之關係或規則。 客戶分群:根據客戶之各種屬性(如性別、職業、收入或教育程度)分析後加以分群。同一群客戶即代表其整體屬性較為類似,依此可進行差異化行銷。 目標式客戶區隔:將所有客戶之單一消費習性加以紀錄、調查,將客戶加以區隔,並以決策樹結構,以客戶各屬性建立區隔模式。 特徵分析:即利用分群分析模型將客戶分群,並進一步分析各客戶群特徵。 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

75 6.3 Application of data mining to CRM
Data Mining of the contact center (2) All information collected is valuable assets, the importance of data mining is very high. Data mining : is use automation or semi-automatic way to carry on trend analysis to a large amount of data, and to seek the meaningful relation or rule to enterprises. Customer segment: Analyzing and clustering the customers according to various kinds of attribute of them(like gender, profession, income and Education degree). The same group of customers means that their entirety attribute is similar. According to this, we can proceed difference marketing. Customer segment by goals: Record and investigate the unity consumption habits of all customers to segment them. And according to customers’ characteristics, using decision tree structure to build segment model. Characteristic (features) analysis: Utilize the cluster analyze model to segment customers, and Step forward to analyze each group’s characteristics. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

76 6.3 資料採礦在顧客關係管理之應用 客服中心之資料採礦(3)
貢獻度分析:即分析客戶對企業貢獻度等級。其乃針對所有客戶的某一種消費屬性加以分析,將客戶加以區隔分類 。 期間分析:依據客戶最近購買日,分析各期間購物者之特徵。 頻率分析:以客戶購買頻率為依據,分析各購物頻率的客戶特徵。 金額分析:以客戶購物金額為依據,分析各購物額之客戶特徵。 終身價值分析:以客戶購物日、購買頻率、購物總金額為依據,歸列其價值評分。 協銷規則分析:分析顧客購物明細,得知顧客購物時趨向同時購買哪些商品,進而主動推薦符合顧客興趣之商品 。 序銷規則分析:尋求顧客於不同購物經驗中購物之先後關係,以使產品行銷定位更正確,並大幅降低行銷分析及廣告費用。 預測分析:依據潛在客戶之各種屬性,透過建立好的客戶區隔模式,進而準確預測未來行銷之顧客歸屬於何種客戶類型之目標。 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

77 6.3 Application of data mining to CRM
Data Mining of the contact center (3) Contribution analysis: It means to analyze customer's grade to enterprise's contribution degree. It is to analyze certain consumption attribute for all customers to segment and cluster customers. Period analysis : According to the purchase recently to analyze every the shopper's characteristic during each period. Frequency analysis : According to the customer buys frequency recently to analyze each shopping frequency of the customer’s characteristic. Amount analysis : According to the customers’ shopping amount to analyze each amount of customer’s characteristic 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

78 6.3 Application of data mining to CRM
Data Mining of the contact center (4) Lifetime value analysis : according to customer’s buying day, frequency and amount to generalize the value grades. Cross-sell rule analysis : Analyzing the detail ledger of customer to know which goods they tends to buying at the same time while doing shopping, to recommend goods conform to customer's interests . Sequential-sell analysis : Seek the priority relation that the customer does shopping in different shopping experience to make products marketing orientation more correct and to slash marketing and advertisement cost Forecast (prediction) analysis : According to various kinds of attribute of the potential customers, through built customer segment model to accurately predict the customer of marketing in future will belong to what kinds customer types. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

79 6.3 資料採礦在顧客關係管理之應用 客服中心之資料採礦(4) 典型應用: 根據消費者瀏覽網頁趨勢,推斷其偏好訊息
根據銷售資料發掘顧客消費習性 從消費者消費及繳費資料預警信用卡呆帳可能 在大量交易資料中探勘產品銷售關聯性 在大量客服資料中發覺話題趨勢 根據歷史審核資料,找尋核發信用卡規則 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

80 6.3 Application of data mining to CRM
Data Mining of the contact center (5) Typical application: According to the tendency of customer surf the webpage to infer their preferences. According to the sale data to Explore customer's consumption habits. Early warning credit card debt from customer’s consume and the data of paying. Product sales affiliation (co-relations) in a large amount of trade data. Discover the hot topics in a large amount of customer service data Look for the authorize rule of credit card according to the history verify data. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

81 6.3 資料採礦在顧客關係管理之應用 資料採礦未來趨勢(1) 彈性提升: 經濟性提升:
由於資料採礦工具大多與取樣資料之相關性高,故資料採礦工具宜具備更高延展性、容納更多屬性與資料維度。另透過前端之智慧型群體區隔功能,可加速產生客戶分類群體,利於企業行銷人員建置大量之資料。 經濟性提升: 資料採礦之知識擷取成果,通常為統計分析之結論,其有效性與可執行性若能更落實於產業需求,形成可採取行動之模型,將可提供投資報酬率,提升資料採礦技術之應用廣度與深度。 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

82 6.3 資料採礦在顧客關係管理之應用 資料採礦未來趨勢(2) 標準化: 整合性:
資料採礦小組(The Data Mining Group)已著手發展「預測模型標記語言」(Predictive Model Markup Language, PMML)。PMML是一種XML格式,以描述常見之預測模型,作為其他資料採礦、商業智慧應用程式直接運用的標準化資料格式 整合性: 為了提昇大規模資料倉儲環境下之資料採礦模型運作效能,部分資料採礦功能乃整合於關聯式資料庫系統之核心功能中,結合關聯式資料庫管理系統之平行運算處理能力,可顯著提昇了資料採礦與查詢的效能。 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬

83 6.3 Application of data mining to CRM
Future trend of data mining Elasticity improving :Because data mining tools are high relation with sample, so the tool should have high expandable property, store more multiple higher-dimension data. Economy improving :The conclusion of analysis can be implemented in industry's demand, offer the rate of returns of investment. Standardization:The Data Mining Group” start to develop PMML (Predictive Model Markup Language) that is a XML standard to describe the common prediction model for other data mining, BI application. Integration:In order to promote the operation efficiency of data mining model, some data mining functions integrate RDBMS to promote query and selecting efficacy. 作者:張力元、姚銀河、侯建良、何佩勳、許芙瑲 總編審:張瑞芬


Download ppt "第六章 資料倉儲與採礦技術 6.1 資料倉儲與採礦定義 6.2 資料採礦之步驟與技術分類 6.3 資料採礦在顧客關係管理之應用"

Similar presentations


Ads by Google