Lluis Belanche + Alfredo Vellido. Intelligent Data Analysis and Data Mining. Data Analysis and Knowledge Discovery

Transcription

1 Lluis Belanche + Alfredo Vellido Intelligent Data Analysis and Data Mining or Data Analysis and Knowledge Discovery a.k.a. Data Mining II

2 Office 319, Omega, BCN EET, office 107, TR 2, Terrassa skype, gtalk: avellido Tels.: , raco.fib.upc.edu/home/assignatura?espai= raco.fib.upc.edu/home/assignatura?espai=270650

3

4 Contents of the course (but who knows) 1. Introduction to DM and its methodologies 2. Visual DM: Exploratory DM through visualization 3. Pattern recognition 1 4. Pattern recognition 2 5. Feature extraction 6. Feature selection 7. Error estimation 8. Linear classifiers, kernels and SVMs 9. Probability in Data Mining 10. Nonlinear Dimensionality Reduction (NLDR) 11. Applications of NLDR: biomed & beyond 12. DM Case studies

5 2013/2014. Alfredo Vellido An Introduction to Mining (1)

6 What is DATA MINING?

7 What is DATA MINING? (1) Data Mining is the process of discovering actionable and meaningful patterns, profiles, and trends by sifting through your data using pattern recognition technologies ( ) is a hot new technology about one of the oldest processes of human endeavour: pattern recognition ( ) It is an iterative process of extracting knowledge from business transactions ( ) DM is the automatic discovery of usable knowledge from your stored data. Jesús Mena: Data Mining your Website (Digital Press, 1999, books.google)

8 What is DATA MINING? (2) Data Mining, by its simplest definition, automates the detection of relevant patterns in a database ( ) For many years, statisticians have manually mined databases ( ) DM uses well established statistical and machine learning techniques to build models that predict customer behaviour. Today, technology automates the mining process, integrates it with commercial data warehouses, and presents it in a relevant way for business users ( ) the leading DM products address the broader business and technical issues, such as their integration into complex IT environments. Berson, Smith, & Thearling: Building Data Mining Applications for CRM (McGraw Hill, 2000)

9 What is DATA MINING? (3) WIKIPEDIA 2005 DIXIT: Data mining has been defined as "The nontrivial extraction of implicit, previously unknown, and potentially useful information from data" (1) and "The science of extracting useful information from large data sets or databases" (2). Although it is usually used in relation to analysis of data, data mining, like artificial intelligence, is an umbrella term and is used with varied meaning in a wide range of contexts. (1) W. Frawley and G. Piatetsky Shapiro and C. Matheus, Knowledge Discovery in Databases: An Overview. AI Magazine, 1992, (2) D. Hand, H. Mannila, P. Smyth: Principles of Data Mining. MIT Press, en.wikipedia.org/wiki/data_mining

10 What is DATA MINING? (4) WIKIPEDIA 06 DIXIT: Data mining (DM), also called Knowledge Discovery in Databases (KDD) or Knowledge Discovery and Data Mining, is the process of automatically searching large volumes of data for patterns such as association rules. It is a fairly recent topic in computer science but applies many older computational techniques from statistics, information retrieval, machine learning and pattern recognition.

11 DAKD,KDD,KDDM In 1996, in the proceedings of the 1st International Conference on KDD, Fayyad gave one of the best known definitions of Knowledge Discovery from Data: The non trivial process of identifying valid, novel, potentially useful, and ultimately understandable patterns in data. KDD quickly gathered strength as an interdisciplinary research field where a combination of advanced techniques from Statistics, Artificial Intelligence, Information Systems, and Visualization are used to tackle knowledge acquisition from large data bases. The term Knowledge Discovery from Data appeared in 1989 referring to the: [...] overall process of finding and interpreting patterns from data, typically interactive and iterative, involving repeated application of specific data mining methods or algorithms and the interpretation of the patterns generated by these algorithms.

12 What is DATA MINING? (6) WIKIPEDIA 08 DIXIT: Data mining is the process of sorting through large amounts of data and picking out relevant information. It is usually used by business intelligence organizations, and financial analysts, but is increasingly being used in the sciences to extract information from the enormous data sets generated by modern experimental and observational methods. It has been described as "the nontrivial extraction of implicit, previously unknown, and potentially useful information from data" and "the science of extracting useful information from large data sets or databases." Data mining in relation to enterprise resource planning is the statistical and logical analysis of large sets of transaction data, looking for patterns that can aid decision making.

13 What is DATA MINING? (7) WIKIPEDIA 10 gave up:

14 What is DATA MINING? (8) but never lose your faith W 13 Data mining (the analysis step of the "Knowledge Discovery in Databases" process, or KDD), an interdisciplinary subfield of computer science, is the computational process of discovering patterns in large data sets involving methods at the intersection of artificial intelligence, machine learning, statistics, and database systems. The overall goal of the data mining process is to extract information from a data set and transform it into an understandable structure for further use. Aside from the raw analysis step, it involves database and data management aspects, data pre processing, model and inference considerations, interestingness metrics, complexity considerations, post processing of discovered structures, visualization, and online updating. The term is a buzzword, and is frequently misused to mean any form of large scale data or information processing (collection, extraction, warehousing, analysis, and statistics) but is also generalized to any kind of computer decision support system...

15 A different (practical) approach to the definition of DM: What to expect from a DM conference September 04: Wessex Institute of Technology (W.I.T.)

16 What to find in a DM conference Sessions 1 & 2: Text Mining Session 3: Web Mining Session 4: Clustering Techniques Session 5: Data Preparation Techniques Session 6 & 7: Applications in Business, Industry and Government Session 8: Customer Relationship Management (CRM) Session 9 & 10: Applications in Science and Engineering

17 What to find in a DM conference (three years later) 2007 Session 1: Categorisation Methods Session 2: Data Preparation Session3: Enterprise InformationSystems Session 4: Clustering Techniques Session 5: National Security Session 6: Data and Text Mining Session 7: Mining Environmental and Geospatial Data Session 8: Applications in Business, Industry and Government

18 IDADM What to find in the dark last few years

19 What to find in the dark last few years Investigative Data Mining For Security And Criminal Detection Jesús Mena Butterworth Heinemann 2003

20 A different conference, a different take IEEE CIDM 2012, Brussels 2012 IEEE Symposium on Computational Intelligence and Data Mining Data mining foundations Novel data mining algorithms in traditional areas (such as classification, regression, clustering, probabilistic modeling, and association analysis) Algorithms for new, structured, data types (chemistry, biology, environment, and other scientific domains) Developing a unifying theory of data mining Mining sequences and sequential data Mining spatial and temporal datasets Mining textual and unstructured datasets High performance implementations of data mining algorithms

21 A different conference, a different take IEEE CIDM 2012, Brussels 2012 IEEE Symposium on Computational Intelligence and Data Mining Mining in targeted application contexts Mining high speed data streams Mining sensor data Distributed data mining and mining multi agent data Mining in networked settings: web, social and computer networks, and online communities Data mining in electronic commerce, such as recommendation, sponsored web search, advertising, and marketing tasks

22 A different conference, a different take IEEE CIDM 2012, Brussels 2012 IEEE Symposium on Computational Intelligence and Data Mining Methodological aspects and the KDD process Data pre processing, data reduction, feature selection, and feature transformation Quality assessment, interestingness analysis, and post processing Statistical foundations for robust and scalable data mining Handling imbalanced data Automating the mining process and other process related issues Dealing with cost sensitive data and loss models Human machine interaction and visual data mining Security, privacy, and data integrity

23 A different conference, a different take IEEE CIDM 2012, Brussels 2012 IEEE Symposium on Computational Intelligence and Data Mining Integrated KDD applications and systems Bioinformatics, computational chemistry, geoinformatics, and other science & engineering disciplines Computational finance, online trading, and analysis of markets Intrusion detection, fraud prevention, and surveillance Healthcare, epidemic modeling, and clinical research Customer relationship management Telecommunications, network and systems management

24 But let s talk money... Starved for ca$h?: ask your TIA

25 The T.I.A. The W Bush years The Total Information Awareness (TIA) program may have been killed by congressional decree, but key elements of the program have survived at other intelligence agencies, according to congressional, federal, and research officials. TIA's goal was to employ data mining to shift through public and private databases to track terrorists, which stirred up fears that the program would be used to spy on millions of innocent Americans. Congressional officials have not disclosed which TIA programs were eliminated and which were retained, but insiders report that TIA's Evidence Extraction and Link Discovery projects, collectively encompassing 18 data mining initiatives, are among the surviving components. Despite the death of TIA, Capitol Hill is still paying for the development of software designed to collect foreign intelligence on terrorists: a $64 million research program run by the Advanced Research and Development Activity (ARDA), which has employed some of the same researchers as TIA, was left untouched by Congress.

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40 What s DATA MINING?: A procedural viewpoint

41 What s DATA MINING?: A historicist viewpoint STATISTICS ESTADÍSTICA DM PATT RECOG KDD ARTIFICIAL INTELLIGENCE EXPERT SYSTEMS MACHINE LEARNING DB MANAGEMENT

42 What s DATA MINING?: A historicist viewpoint STATISTICS ESTADÍSTICA KDD ADVANCED PROBABILISTIC MODELS Probabilistic Models ARTIFICIAL INTELLIGENCE MACHINE LEARNING OTHERS Algor. Devel. Bio-plausible Models

43 DATA MINING as a methodology

44 CRISP: a DM methodology CRoss Industry Standard Process for Data Mining: neutral methodology from the point of view of industry, tool and application (free &nonproprietary) Pete Chapman, Randy Kerber (NCR); Julian Clinton, Thomas Khabaza, Colin Shearer (SPSS), Thomas Reinartz, Rüdiger Wirth (DaimlerChrysler) CRISP DM was conceived in 1996 DaimlerChrysler: leaders in industrial application, SPSS: leaders in product development (Clementine, 1994), NCR: owners of large (huge!) databases (Teradata) Financed by the EU. Version 1.0 released officially in 1999

45 CRISP: Hierarchic structure of the methodology

46 CRISP: Description of phases Problem understanding: study of targets and requirements form the business/problem viewpoint. Defining it as a DM problem. Data understanding: data recolection; getting to know the data, trying to detect both quality problems and interesting features. Data preparation: Preparing the data set to be modelled, starting from raw data. This is an iterative and exploratory process. Selection of files, tables, variables, record samples plus data cleaning. Modelling: Data analysis using modelling techniques of a sort that are suitable for the problem at hand. Includes fiddling with the models, tuning their parameters, etc. Evaluation: All previous steps must be evaluated as whole (as a unitary process), and we must decide whether deliverables so far meet the DM challenge. Implementation: All the knowledge aquired to this point must be organized and presented to the client in a usable form. We must define, together with this client, a protocol to reliably deploy the DM findings.

47 CRISP: The virtuous loop of methodology phases

48 Use of DM methodologies (2004) Enterprise MinerTM: SEMMA The acronym SEMMA Sample, Explore, Modify, Model, Assess refers to the core process of conducting data mining. Beginning with a statistically representative sample of your data, SEMMA makes it easy to apply exploratory statistical and visualization techniques, select and transform the most significant predictive variables, model the variables to predict outcomes, and confirm a model's accuracy.

49 Use of DM methodologies ( )

50 CRISP: Phases: Problem understanding PROBLEM UNDERSTANDING DATA UNDERST ING DATA PREPARATION MODELLING EVALUATION IMPLEMEN TATION DETERMINE PROBLEM GOAL BACKGROUND PROBLEM GOALS SUCCESS CRITERIA ASSESS SITUATION INVENTORY RESOURCES REQUERIMS. ASSUMPTIONS LIMITATIONS RISKS CONTINGEN. TERMINOLOG. COSTS & BENEFITS DETERMINE DM GOALS GOALS DM SUCCESS CRITERIA DM PRODUCE PROJECT PLAN PROJECT PLAN INITIAL SELECTION OF TOOLS

51 DM applicationareas ( 06 > 09)

52 DM application areas ( 09 > 10)

53 DM application areas ( 10 > 11)

54 CRISP: Phases: Data understanding PROBLEM UNDERSTANDING DATA UNDERST ING DATA PREPARATION MODELLING EVALUATION IMPLEMEN TATION OBTAIN INITIAL DATA INITIAL DATA REPORT DESCRIPTION DATA EXPLORATION DATA VERIFICATION QUALITY DATA DATA DESCRIPTIVE REPORT DATA EXPLORATION REPORT DATA QUALITY REPORT