Data mining for shopping centres - cData mining for shopping centres - customer knowledge-management fraustomer knowledge-management fra

meworkmework

授課教師 : 許素華博士

學生 : S92660005 黃永智

S92660014 呂曉康

S92660017 李峻賢

日期 : 2004/03/29

Dennis, C., Marsland, D., Cockett, T.(2001)  Journal of Knowledge Management. . Vol. 5, Iss. 4;  pp. 368-374

2

AgendaAgenda

Introduction Exploratory Study Results Models of Relative Spend Discussion and Conclusion About K-Means

3

Introduction(1/2)Introduction(1/2)

Knowledge is a fundamental factor behind an enterprise's success Management using knowledge-based computer systems anManagement using knowledge-based computer systems an

d networks d networks Management of intellectual (human) capitalManagement of intellectual (human) capital All knowledge activities affecting success All knowledge activities affecting success

Richards et al. (1998) argue that success is founded on "a continuous dialogue with users, leading to a real understanding".

For retailers the key ... is to establish data warehouses to improve and manage customer relationships (Teresko, 1999)

4

Introduction(2/2)Introduction(2/2)

Incorporating data mining and customer database aspects within a framework of knowledge management can help increase knowledge value.

Sharing information Loyalty schemes The objective of retail data mining schemes

has been to identify subgroups Shopping and Service motivations Shopping and Service motivations

5

Exploratory Study Exploratory Study

The results are from a survey of 287 respondents at six shopping centres

Determine which specific attributes of shopping centres were most associated with spend for subgroups of shoppers

Convenience sample –(weekdays, 10.30am to 3.30pm )

Unstructured Interviews Unstructured Interviews Least squares regression

ResultsResults

7

Conventional demographics(1/6)Conventional demographics(1/6)

Females vs males The significant attributes for females were grouped around tThe significant attributes for females were grouped around t

wo factors:wo factors: Shopping: "selection of merchandise"Shopping: "selection of merchandise" Experience: "friendly atmosphere" Experience: "friendly atmosphere"

8

Conventional demographics(2/6)Conventional demographics(2/6)

Upper vs Lower socio-economic groups ABC1 (managerial, administrative, professional, supervisoABC1 (managerial, administrative, professional, superviso

ry and clerical)ry and clerical) C2DE (manual workers and pensioners) C2DE (manual workers and pensioners)

9

Conventional demographics(3/6)Conventional demographics(3/6)

Higher vs Lower income groups

10

Conventional demographics(4/6)Conventional demographics(4/6)

Older vs Younger shoppers

11

Conventional demographics(5/6)Conventional demographics(5/6)

Shoppers travelling by car vs Public transport

12

Conventional demographics(6/6)Conventional demographics(6/6)

Service importance vs Shops importance

13

Cluster analysisCluster analysis

Shoppers motivated by the "importance" of "shops" vs "service“ A cluster analysis (SPSS A cluster analysis (SPSS

K-means) based on "impK-means) based on "importance" scores has identortance" scores has identified distinct subgroups sified distinct subgroups sharing particular needs oharing particular needs or wants. r wants.

14

Compare Service & Shops GroupsCompare Service & Shops Groups

“Service" shoppers were in a slightly higher characters then “shops” shoppers gropus Socio-economic group (63 % ABC1s vs. 59 %) Socio-economic group (63 % ABC1s vs. 59 %) Income (60 % Income (60 % ￡￡ 20,000 per year + vs. 53 %)20,000 per year + vs. 53 %) Age (42 percent 45 + vs. 33 percent)Age (42 percent 45 + vs. 33 percent) Traveled by car (90 percent vs. 52 percent) Traveled by car (90 percent vs. 52 percent)

15

Models of Relative Spend Models of Relative Spend

for "shops": 11 Spend = 19.4 + 0.70 X Attractiveness -0.21 X Distance.

16

Relationship Between Attractiveness & Sales Relationship Between Attractiveness & Sales Turnover Turnover

17

Discussion and Conclusion Discussion and Conclusion

Target high-spending "service" shoppers. Increase their spend by 10%, equivalent to over a 3% rise iIncrease their spend by 10%, equivalent to over a 3% rise i

n total sales. n total sales. Local loyalty cards are applicable and cost-effective f

or cities and regional shopping centres Car park membership scheme for in-town centresCar park membership scheme for in-town centres

Knowledge management network between retailers and the centre would be a further stage

Most successful shopping centres are those where “Active marketing" and “Proactive management" are a feature

18

About K-Means

The K-Means AlgorithmThe K-Means Algorithm

1. Choose a value for K, the total number of clusters.

2. Randomly choose K points as cluster centers.

3. Assign the remaining instances to their closest cluster center.

4. Calculate a new cluster center for each cluster.

5. Repeat steps 3-5 until the cluster centers do not change.

An Example Using K-Means

Table 3.6 • K-Means Input Values

Instance X Y 1 1.0 1.5 2 1.0 4.5 3 2.0 1.5 4 2.0 3.5 5 3.0 2.5 6 5.0 6.0

Figure 3.6 A coordinate mapping of the data in Table 3.6

0

1

2

3

4

5

6

7

0 1 2 3 4 5 6

f(x)

x

Table 3.7 • Several Applications of the K-Means Algorithm (K = 2)

Outcome Cluster Centers Cluster Points Squared Error

1 (2.67,4.67) 2, 4, 6 14.50

(2.00,1.83) 1, 3, 5

2 (1.5,1.5) 1, 3 15.94

(2.75,4.125) 2, 4, 5, 6

3 (1.8,2.7) 1, 2, 3, 4, 5 9.60

(5,6) 6

Figure 3.7 A K-Means clustering of the data in Table 3.6 (K = 2)

0

1

2

3

4

5

6

7

0 1 2 3 4 5 6

x

f(x)

General ConsiderationsGeneral Considerations

Requires real-valued data. We must select the number of clusters present in

the data. Works best when the clusters in the data are

of approximately equal size. Attribute significance cannot be determined. Lacks explanation capabilities.

26

簡報完畢 敬請指導