Target Node
: B_NUMBER
Date Column
: EVT_DATE
Links Type
: Directed
Use a Weight Column:
USG_DURATION
Choose
Next
In the
Post-Processings
screen, allow
Community Detection
and
Mega-hub detection
Choose
Node Pairing
page and select
+
to add.
In the
Pairing Definition
screen, enter the parameters as follows:
a. First Graph: Churners_Network
b. Second Graph: Newcomers_Network
c. Output Graph Name: Social Network Comparison
d. Set
Keep Top N
to a desired value for better performance
e. Enable Minimum Common Neighbours Ratio and enter a value between typically between 0.5
and 1
Pairing Type : Ratio
Ensure that the checkbox for Weighted Ratio and Include Common Neighbours Count Graph
are checked
Click Ok
Choose Next and skip Add Identifiers Conversion Data Set.
Choose Next and skip Add Descriptive Attributes
Choose Generate to construct the social graph
To verify and to analyze the model generated click on "Nodes Display".
a. Choose Graph "Social_Network_Comparison_Ratio" and choose a phone number and choose
display node. The graph shows similar phone numbers
b. The phone numbers can be individually analyzed in Churners_Network and
Newcomers_Network.
c. Chose Reports -> Compare to compare the two indivudal graphs.
After evaluating the model, select
Run
Choose
Apply Model
and use following setting in
Applying the Model
screen:
Application Data Set
Data Type: Data Base
Folder: <HANA connection>
Data: DM_TELCO_LIST_OF_NEWCOMERS
In Define Mapping, map the following:
A_NUMBER -> PHONE_NUMBER
kxcomIndex -> any arbitrary value(since we are using database option, this is not significant)
Generation options
Generate: Neighbors Mode
In Advanced Apply Settings,
Check Nodes Neighborhood for Social Network Comparison Ratio.
Uncheck the checkbox for all other graphs
Mode: Apply
Results Generated by the Model
Data Type: Data Base
