9654467111 Call Girls In Mahipalpur Women Seeking Men
Cluster analysis
1.
2. NOW FOCUS ON BUSINESS KEY PARAMETERS
Price paid
Discount
Sales
Unit sold
ANALYSIS SOFTWARE SAS
3. DESCRIBTION OF THE PROJECT:
The project aims to construct a cluster analysis model
on the basis of a given set of variables.
Cluster Analysis means a techniques used to classify
cases into groups that areRelatively Homogeneous within themselves &
Heterogeneous between each other.
Heterogeneity & Homogeneity are measured on the
basis of given set of variables.
6. By Factor Analysis we consider those factors which explain 95%
of total variation with data.
Med_Actual_price_12
AVG_Actual_price_12
TOTAL_LINE_ITEM_QTY_12
AVG_LINE_ITEM_QTY_12
AVG_DOLLAR_DISCOUNT_12
SIZE_QTY
SIZE_DOLLAR_DISC
RATIO_DISC_SALES
Med_LINE_ITEM_QTY_12
Med_DOLLAR_DISCOUNT_12
MONTH_SINCE_LAST_TRANSACTION
Count
Sum_value_12
7. For these selected factors we Standardized the
variables to make them unit free since the units of
measurement of different variables are different.
8. Next step is cluster building using “K-means clustering”
which satisfies the following criterion: Individual R-Squared > 0.25
Overall R-Squared > 0.5
Approximate Expected Overall R-Squared > 0.3
(This is R-Squared if there was no Multicollineraity)
The difference Between Overall R-Squared and
Approximate Expected Overall R-Squared should
not be greater than 0.2
RMS Standard Deviation < 1.4
Distance Between Cluster Centroids > 1.4
13. CLUSTER VALIDATION
NO OF CLUSTER
1
2
CLUSTER NAME
EXTREMELY VALUE SEEKER
EXTREMELY DISCOUNT
MINDED
DISCOUNT ORIENTED BUT
ALSO VALUE SEEKER
3
FREQUENCY
76
83
78
4
VALUE SEEKER BUT DON'T LET
DISCOUNT
116
5
MIXED ORIENTATION
245
6
VALUE SEEKER
109
18. VALUE SEEKER BUT DON'T LET DISCOUNT
1%
17%
15%
med_Actual_price_12
med_DOLLAR_DISCOUNT_12
67%
AVG_DOLLAR_DISCOUNT_12
MONTH_SINCE_LAST_TRANSACTION
19. DISCOUNT ORIENTED BUT ALSO VALUE SEEKER
0%
36%
36%
med_Actual_price_12
med_DOLLAR_DISCOUNT_12
AVG_DOLLAR_DISCOUNT_12
MONTH_SINCE_LAST_TRANSACTION
28%
21. Based on our analysis we can conclude that , we get six
different cluster ‘s of customer orientation.
In first cluster customer extremely oriented towards
value for money.
In second cluster we saw just extreme opposite pattern
of first cluster , here customer preferred discount
more.
In this cluster customer’s are drawn towards discount
but they also look for value.
22. In the forth cluster customer are value seeker but they
also look for discount , here we also found out the
significance of another variable(Month since last
transaction), which was insignificant in other five
cluster’s. Although this variable account for only one
percent in this cluster . We can’t ignore single
customer.
In this cluster customer preference ix mixed between
three variable.
Last but not the least this cluster full with purely value
seeker customer.
23. Based on so far cluster profiling this software
company can make different strategies for
different customer group in different cluster.
Thank You