This preview shows page 1. Sign up to view the full content.
Unformatted text preview: CS345 Data Mining
Introductions What Is It? Cultures of Data Mining
1 Course Staff
x Instructors: Anand Rajaraman Jeff Ullman Robbie Yan x TA: 2 Requirements
x Homework (Gradiance and other) 20% x Project 40% x Final Exam 40% Gradiance class code BB8F698B 3 Project
x Software implementation related to course subject matter. x Should involve an original component or experiment. x We will provide some databases to mine; others are OK. 4 Team Projects
x Working in pairs OK, but ...
1. We will expect more from a pair than from an individual. 2. The effort should be roughly evenly distributed. 5 What is Data Mining?
x Discovery of useful, possibly unexpected, patterns in data. x Subsidiary issues: E.g., age = 150. Data cleansing: detection of bogus data. Visualization: something better than megabyte files of output. Warehousing of data (for retrieval).
6 Typical Kinds of Patterns
1. Decision trees: succinct ways to classify by testing properties. 2. Clusters: another succinct classification by similarity of properties. 3. Bayes, hiddenMarkov, and other statistical models, frequentitemsets: expose important associations within data.
7 Example: Clusters
x x x x x x x x xx x x x x x x x x xx x x x x x x x xx x x x x x x x x x x x x 8 Example: Frequent Itemsets
x A common marketing problem: examine what people buy together to discover patterns.
1. What pairs of items are unusually often found together at Safeway checkout? Answer: diapers and beer. 1. What books are likely to be bought by the same Amazon customer?
9 Applications (Among Many)
x Intelligencegathering. x Web Analysis. x Marketing. PageRank. Total Information Awareness. Run a sale on diapers; raise the price of beer.
x Databases: concentrate on largescale (nonmainmemory) data. x AI (machinelearning): concentrate on complex methods, small data. x Statistics: concentrate on inferring models. 11 x To a database person, datamining is a powerful form of analytic processing queries that examine large amounts of data. x To a statistician, datamining is the inference of models. Result is the data that answers the query. Models vs. Analytic Processing Result is the parameters of the model.
12 (Way too Simple) Example
x Given a billion numbers, a DB person might compute their average. x A statistician might fit the billion points to the best Gaussian distribution and report the mean and standard deviation. 13 Meaningfulness of Answers
x A big risk when data mining is that you will "discover" patterns that are meaningless. x Statisticians call it Bonferroni's principle: (roughly) if you look in more places for interesting patterns than your amount of data will support, you are bound to find crap.
x A big objection to TIA was that it was looking for so many vague connections that it was sure to find things that were bogus and thus violate innocents' privacy. x The Rhine Paradox: a great example of how not to conduct scientific research.
15 Rhine Paradox (1)
x David Rhine was a parapsychologist in the 1950's who hypothesized that some people had ExtraSensory Perception. x He devised an experiment where subjects were asked to guess 10 hidden cards red or blue. x He discovered that almost 1 in 1000 had ESP they were able to get all 10 right!
16 Rhine Paradox (2)
x He told these people they had ESP and called them in for another test of the same type. x Alas, he discovered that almost all of them had lost their ESP. x What did he conclude? Answer on next slide. 17 Rhine Paradox (3)
x He concluded that you shouldn't tell people they have ESP; it causes them to lose it. 18 A Concrete Example
x This example illustrates a problem with intelligencegathering. x Suppose we believe that certain groups of evildoers are meeting occasionally in hotels to plot doing evil. x We want to find people who at least twice have stayed at the same hotel on the same day.
19 The Details
x 109 people being tracked. x 1000 days. x Each person stays in a hotel 1% of the time (10 days out of 1000). x Hotels hold 100 people (so 105 hotels). x If everyone behaves randomly (I.e., no evildoers) will the data mining detect anything suspicious?
20 Calculations (1)
x Probability that persons p and q will be at the same hotel on day d : x Probability that p and q will be at the same hotel on two given days: x Pairs of days: 5*105. 109 * 109 = 1018. 1/100 * 1/100 * 105 = 109. 21 Calculations (2)
x Probability that p and q will be at the same hotel on some two days: x Pairs of people: 5*1017. 5*105 * 1018 = 5*1013. x Expected number of suspicious pairs of people: 5*1017 * 5*1013 = 250,000.
x Suppose there are (say) 10 pairs of evil doers who definitely stayed at the same hotel twice. x Analysts have to sift through 250,010 candidates to find the 10 real cases. Not gonna happen. But how can we improve the scheme?
View Full Document
This document was uploaded on 03/04/2012.
- Fall '09
- Data Mining