Data Science Course in Pune

1. Market Basket Analysis AssociationRules Relationship Mining Affinity Analysis © 2013 ExcelR Solutions. All RightsReserved

2. Market BasketAnalysis • Large number of transaction records through data collected using bar-codescanners • Each record = All items purchased on a single purchasetransaction © 2013 ExcelR Solutions. All RightsReserved

3. AssociationRules • What item goes withwhat • Are certain groups of items consistently purchasedtogether • What business strategies will you device with thisknowledge © 2013 ExcelR Solutions. All RightsReserved

4. AssociationRules • Products shelf placement – a specific product besideanother • Selling of prominent shelves – SlottingFees • Stocking – Supply ChainManagement • Price Bundling – Combo offers.How? • Source:http://www.economist.com/news/business/21654601-supplier-rebates-are-heart-some-supermarket-chains-woes-buying-up-shelves • https://en.wikipedia.org/wiki/Association_rule_learning © 2013 ExcelR Solutions. All RightsReserved

5. A store sells accessories for cellular phones runs a promotion on faceplates • OFFER! • Buy multiple faceplates from a choiceof • 6 different colors & getdiscount • How would you help store managers device strategy to become more profitable Association Rules – Cell phonefaceplates © 2013 ExcelR Solutions. All RightsReserved

6. ListFormat Binary MatrixFormat Association Rules – Cell phonefaceplates • Association Rules are probabilistic “if-then”statements • 2 Main Ideas: • Examine all possible “if-then” ruleformats • Select rules, which indicates truedependence © 2013 ExcelR Solutions. All RightsReserved

7. Rules for { Red, White, Green} • Problem • Many rules arepossible • How to select the TRUE/GOOD rulesfrom all generatedrules? Association Rules – Cell phonefaceplates © 2013 ExcelR Solutions. All RightsReserved

8. “IF” part = Antecedent =A • “THEN” part = Consequent =C • If {Red, White} then{Green} • If Red & White phone faceplates are purchased, then Green faceplate is purchased • Antecedent: Red &White • Consequent:Green Association Rules –Terminology © 2013 ExcelR Solutions. All RightsReserved

9. Association Rules – Performance Measures 1 Support 2 Confidence 3 Lift © 2013 ExcelR Solutions. All RightsReserved

10. Consider only combinations that occur with higher frequency in thedatabase • Support is the criterion based onfrequency • Percentage / Number of transactions in which IF/Antecedent & THEN / Consequent appear in thedata Association Rules –Support 1 Support Mathematically: # transactions in which A & C appeartogether Total no. oftransactions © 2013 ExcelR Solutions. All RightsReserved

11. Support -Calculation • What is the support for “if White thenBlue”? • What is the support for “if Blue then White”? 1. 4 2. 40% 3. 2 4. 90% 1. 4 2. 40% 3. 2 4. 90% © 2013 ExcelR Solutions. All RightsReserved

12. Generating all possible rules is exponential in the number of distinctitems • Solution: • Frequent item sets using AprioriAlgorithm Support -Problem © 2013 ExcelR Solutions. All RightsReserved

13. AprioriAlgorithm For kproducts: 1 Set minimum supportcriteria Generate list of one-item sets that meet thesupport 2 criterion Use list of one-item sets to generate list of two-itemsets 3 that meet supportcriterion Use list of two-item sets to generate list of three-itemsets 4 that meet supportcriterion 5 Continue up through k-itemsets © 2013 ExcelR Solutions. All RightsReserved

14. Support – Criterion =2 Create rules from frequent item setsonly © 2013 ExcelR Solutions. All RightsReserved

15. Rules for { Red, White, Green} Support CriterionExample © 2013 ExcelR Solutions. All RightsReserved

16. Percentage of If/Antecedent transactions that also have the Then/Consequent itemset Association Rules –Confidence Mathematically: P (Consequent | Antecedent) = P(C & A) /P(A) 2 Confidence # transactions in which A & C appeartogether # transactions withA © 2013 ExcelR Solutions. All RightsReserved

17. Confidence -Calculation • What is theconfidence • for “if White thenBlue”? • What is theconfidence • for “if Blue thenWhite”? 1. 4/5 2. 5/8 3. 5/4 4. 4/8 1. 4/5 2. 5/8 3. 5/4 4. 4/8 © 2013 ExcelR Solutions. All RightsReserved

18. If antecedent and consequenthave: • HighSupport => High / BiasedConfidence Confidence -Weakness © 2013 ExcelR Solutions. All RightsReserved

19. Confidence / Benchmark confidence Benchmark assumes independence between antecedent & consequent: Association Rules – LiftRatio Benchmarkconfidence P(C|A) = P(C & A) / P(A) = P(C) X P(A) /P(A) =P(C) 3 LiftRatio # transactions with consequent itemsets # transactions indatabase © 2013 ExcelR Solutions. All RightsReserved

20. Lift > 1 indicates a rule that is useful in finding consequent item sets • The rule above is much better than selecting randomtransactions InterpretingLift © 2013 ExcelR Solutions. All RightsReserved

21. Lift -Calculation • What is the Lift for “if White thenBlue”? • 1. 4/8 • 2. 5/10 • 3. 4/5 • 4. 1 © 2013 ExcelR Solutions. All RightsReserved

22. Generate all rules that meet specified Support &Confidence • Find frequent item sets based on Support specified by applying minimum supportcutoff • From these item sets, generate rules with defined Confidence. By filtering remaining rules select only thosewith highConfidence Rules selection process © 2013 ExcelR Solutions. All RightsReserved

23. Rules © 2013 ExcelR Solutions. All RightsReserved

24. Random data can generateapparently interesting associationrules • More the rules you produce, greater the danger • Rules based on large numbers of records • are less subject to thisdanger Alarming! © 2013 ExcelR Solutions. All RightsReserved

25. Profusion ofrules © 2013 ExcelR Solutions. All RightsReserved

26. What if Product & Stores are selected as a tuple foranalysis? Applications • What if crimes in different geographies for eachweek is known? Narcotics Public Peace Violation Battery Assault Narcotics Robbery © 2013 ExcelR Solutions. All RightsReserved

27. How can you use the information if you know aboutthe • purchase history of customers in a specificgeography? • Supermarket database has 100,000 POStransactions Recap with anexample • 2000 transactions include both Strepsils & OrangeJuice • 800 of the above 2000 include Souppurchases © 2013 ExcelR Solutions. All RightsReserved

28. What is the support for rule “IF (Orange Juice & Strepsils) are purchased THEN (Soup) is purchased on the sametrip”? • 1. 0.8% • 2. 2% • 3. 40% Recap with anexample • What is the confidence for rule “IF (Orange Juice & Strepsils) arepurchased • THEN (Soup) is purchased on the sametrip”? • 1. 0.8% • 2. 2% • 3. 40% © 2013 ExcelR Solutions. All RightsReserved

29. What is the lift ratio for rule “IF (Orange Juice & Strepsils) are purchased THEN (Soup) is purchased on the sametrip”? Recap with anexample © 2013 ExcelR Solutions. All RightsReserved

30. ITIS • If person X has taken “Data Mining Unsupervised” training in 1st Quarter, Person X has alsotaken “Data Mining training in 2nd Supervised” Quarter Sequential PatternMining • Based on the statement • above, recommend “Data NOT Mining Supervised” training to those who have enrolled for “Data MiningUnsupervised” Purchases / events occur at the sametime © 2013 ExcelR Solutions. All RightsReserved

31. Look for temporalpatterns • Order/sequence of a & b matters for a rule “b followsa” • However, what happens in between a & b doesn’tmatter • In phone faceplates dataset: • Association among items, which were bought withinthe same week werediscovered • How about finding what they would buy next week or the week after, if they had bought ‘x’ in thisweek? Association Rules vs. Sequential PatternMining © 2013 ExcelR Solutions. All RightsReserved

32. Identify the appropriateBasket Applications • Identify popular taxiroutes • Sequential pattern from GPS tracks; spatiotemporal records of taxi trajectories • First cluster collocatedcustomers © 2013 ExcelR Solutions. All RightsReserved

33. www.excelr.com surya@excelr.com +91 9880913504 ExcelR - Data Science, Data Analytics Course Training in Pune Address: 102, 1st Floor, Phase II, Prachi Residency Opposite to KapilMalhar, Baner Rd, Baner, Pune,Maharashtra 411046 Hour: Mon- Sat 07AM – 11PM Established in Year: 2013 CONTACT US THANK YOU © 2013 ExcelR Solutions. All RightsReserved