Presentation is loading. Please wait.

Presentation is loading. Please wait.

N. Kumar, Asst. Professor of Marketing Database Marketing Cluster Analysis.

Similar presentations


Presentation on theme: "N. Kumar, Asst. Professor of Marketing Database Marketing Cluster Analysis."— Presentation transcript:

1 N. Kumar, Asst. Professor of Marketing Database Marketing Cluster Analysis

2 2 Agenda Discussion of the first Assignment Motivation for conducting Cluster Analysis Benefit Segmentation Cluster Analysis Basic Concepts Hierarchical/Non- Hierarchical Clustering Implementation in SAS and interpreting the output

3 N. Kumar, Asst. Professor of Marketing Voter Profiling What are the different voting segments out there? What do they want to hear i.e. issues they care about? What should I say?

4 N. Kumar, Asst. Professor of Marketing Ad Campaign How many customer segments are there? How many do I want to target? How should I target – what message should I communicate to each segment?

5 N. Kumar, Asst. Professor of Marketing Promotional Strategies Coupon Drops – who should they be targeted at? Catalog Example – should the catalog be accompanied with a $5 coupon or a $10 coupon or no coupon?

6 N. Kumar, Asst. Professor of Marketing What is Cluster Analysis? Cluster Analysis is a technique for combining observations into groups or clusters such that: Each group is homogenous with respect to certain characteristics (that you specify) Each group is different from the other groups with respect to the same characteristics

7 N. Kumar, Asst. Professor of Marketing Data ConsumerIncome ($ 1000s)Education (years) 155 266 31514 41615 52519 63020

8 N. Kumar, Asst. Professor of Marketing Geometrical View of Cluster Analysis Education Income

9 N. Kumar, Asst. Professor of Marketing Similarity Measures Why are consumers 1 and 2 similar? Distance(1,2) = (5-6) 2 + (5-6) 2 More generally, if there are p variables: Distance(i,j) =  (x ik - x jk ) 2

10 N. Kumar, Asst. Professor of Marketing Similarity Matrix C1C2C3C4C5C6 C102181221625821 C220145181557745 C318114502136250 C422118120106212 C5625557136106026 C6821745250212260

11 N. Kumar, Asst. Professor of Marketing Clustering Techniques Hierarchical Clustering Non-Hierarchical Clustering

12 N. Kumar, Asst. Professor of Marketing Hierarchical Clustering Distance(1,2) = 2 = Distance(3,4) Say, we group 1 and 2 together and leave the others as is How do we compute the distance between a group that has two (or more) members and the others?

13 N. Kumar, Asst. Professor of Marketing Hierarchical Clustering Algorithms Centroid Method Nearest-Neighbor or Single-Linkage Farthest-Neighbor or Complete-Linkage Average-Linkage Ward’s Method

14 N. Kumar, Asst. Professor of Marketing Centroid Method Each group is replaced by an average consumer Cluster 1 – average income = 5.5 and average education = 5.5

15 N. Kumar, Asst. Professor of Marketing Data for Five Clusters ClusterMembersIncomeEducation 1C1&C25.5 2C31514 3C41615 4C52520 5C63019

16 N. Kumar, Asst. Professor of Marketing Similarity Matrix C1&C2C3C4C5C6 C1&C20 C3162.50 C4200.520 C5590.5135.961060 C6782.5250212260

17 N. Kumar, Asst. Professor of Marketing Data for Four Clusters ClusterMembersIncomeEducation 1C1&C25.5 2C3&C415.514.5 3C52520 4C63019

18 N. Kumar, Asst. Professor of Marketing Similarity Matrix C1&C2C3&C4C5C6 C1&C20 C3&C41810 C5590120.50 C6782.5230.5260

19 N. Kumar, Asst. Professor of Marketing Data for Three Clusters ClusterMembersIncomeEducation 1C1&C25.5 2C3&C415.514.5 3C5&C627.519.5

20 N. Kumar, Asst. Professor of Marketing Similarity Matrix C1&C2C3&C4C5&C6 C1&C20 C3&C41810 C5&C66801690

21 N. Kumar, Asst. Professor of Marketing Dendogram for the Data C1C2C3C4C5C6

22 N. Kumar, Asst. Professor of Marketing Single Linkage First Cluster is formed in the same fashion Distance between Cluster 1 comprising of customers 1 and 2 and customer 3 is the minimum of Distance(1,3) = 181 and Distance(2,3) = 145

23 N. Kumar, Asst. Professor of Marketing Similarity Matrix C1&C2C3C4C5C6 C1&C20 C31450 C418120 C55571361060 C6745250212260

24 N. Kumar, Asst. Professor of Marketing Complete Linkage Distance between Cluster 1 comprising of customers 1 and 2 and customer 3 is the maximum of Distance(1,3) = 181 and Distance(2,3) = 145

25 N. Kumar, Asst. Professor of Marketing Similarity Matrix C1&C2C3C4C5C6 C1&C20 C31810 C422120 C56251361060 C6821250212260

26 N. Kumar, Asst. Professor of Marketing Average Linkage Distance between Cluster 1 comprising of customers 1 and 2 and customer 3 is the average of Distance(1,3) = 181 and Distance(2,3) = 145

27 N. Kumar, Asst. Professor of Marketing Similarity Matrix C1&C2C3C4C5C6 C1&C20 C31630 C420120 C55911361060 C6783250212260

28 N. Kumar, Asst. Professor of Marketing Ward’s Method Does not compute distance between clusters Forms clusters by maximizing within- cluster homogeneity or minimizing error sum of squares (ESS) ESS for cluster with two observations (say, C1 and C2) = (5-5.5) 2 + (6-5.5) 2 + (5-5.5) 2 + (6-5.5) 2

29 N. Kumar, Asst. Professor of Marketing Ward’s Method CL1CL2CL3CL4CL5ESS 1C1,C2C3C4C5C61 2C1,C3C2C4C5C690.5 3C1,C4C2C3C5C6110.5 4C1,C5C2C3C4C6312.5 5C1,C6C2C3C4C5410.5 6C2,C3C1C4C5C672.5 7C2,C4C1C3C5C690.5

30 N. Kumar, Asst. Professor of Marketing Non-Hierarchical Clustering Data are grouped into K clusters Requires a priori knowledge of K

31 N. Kumar, Asst. Professor of Marketing Basic Steps in Non-Hierarchical Clustering Select K initial cluster centroids Assign each observation to the cluster to which it is closest Reassign or reallocate each observation to one of the K clusters according to a pre-determined stopping rule Stop if there is no reallocation Approaches differ in Step 1 and/or step 3

32 N. Kumar, Asst. Professor of Marketing Algorithm I Selects first K observations as cluster centers

33 N. Kumar, Asst. Professor of Marketing Initial Cluster Centroids VariableCL1CL2CL3 Income5615 Education5614

34 N. Kumar, Asst. Professor of Marketing Initial Assignment Distance from C1 Distance from C2 Distance from C3 Assigned to CL C1021811 C2201452 C318114503 C422118123 C56255571363 C68217452503

35 N. Kumar, Asst. Professor of Marketing New Cluster Centroids VariableCL1CL2CL3 Income5621.5 Education5617

36 N. Kumar, Asst. Professor of Marketing Distance Matrix Distance from CL1 Distance from CL2 Distance from CL3 Previous Assignment Current Assignment C102416.1511 C220316.2522 C318114551.2533 C422118134.2533 C562555721.2533 C682199076.2533

37 N. Kumar, Asst. Professor of Marketing Algorithm II Differs from Algorithm I in how the initial seeds are modified As before first K observations are selected as the initial cluster seeds A seed that is a candidate for replacement is from one of the two seeds that are closest to each other An observation qualifies to replace one of the two candidates if the distance between the seeds is less than the distance between the observation and the closest seed

38 N. Kumar, Asst. Professor of Marketing Algorithm II …contd. C1, C2 and C3 are the initial seeds The smallest distance between the seeds is between C1 and C2 Observation C4 does not qualify as a replacement as Distance(C1,C2) > Distance(C4 and the nearest seed C3) Observation C5 does qualify as a replacement as Distance(C1,C2) < Distance(C5 and the nearest seed C3): replace C2 with C5

39 N. Kumar, Asst. Professor of Marketing Initial Assignment Distance from C1 Distance from C2 Distance from C3 Assigned to CL C101816251 C221455571 C318101362 C422121062 C562513603 C6821250263

40 N. Kumar, Asst. Professor of Marketing New Cluster Centroids VariableCL1CL2CL3 Income5.515.527.5 Education5.514.519.5

41 N. Kumar, Asst. Professor of Marketing Distance Matrix Distance from CL1 Distance from CL2 Distance from CL3 Previous Assignment Current Assignment C10.5200.5716.511 C20.5162.5644.511 C3162.50.5186.522 C4200.50.5152.522 C5590.5120.56.533 C6600.50230.56.533

42 N. Kumar, Asst. Professor of Marketing Hierarchical vs. Non-Hierarchical Clustering Hierarchical clustering does not require a priori knowledge of the number of clusters Assignments are static Use hierarchical clustering for exploratory purposes Non-Hierarchical Methods can be viewed as a complementary rather than a competing method

43 N. Kumar, Asst. Professor of Marketing Voter Profiling Survey of voters concerns may help us group customers with similar concerns – perhaps they all live in a certain area? Target ads/mailings with customized messages

44 N. Kumar, Asst. Professor of Marketing Ad Campaign Use attitudinal data to segment customers Target message appropriately

45 N. Kumar, Asst. Professor of Marketing Promotional Strategies Use transaction data to group customers into those that are more prone to purchasing the product on deal Give a stronger incentive to the price sensitive segment


Download ppt "N. Kumar, Asst. Professor of Marketing Database Marketing Cluster Analysis."

Similar presentations


Ads by Google