CTS FWR5-3105SFP-CW-AC-DR Series, Руководство пользователя

Страница: 37 / 90

Results Generated by the Model
Data Type: Text Files
Folder: local folder
Data: result_class_pref_ind_summary_active_adv.txt
Select Apply
After reviewing model results, choose Next.
Choose Save/Export and choose Save Model
Saving the Model
Model Name: default
Description: description of model
Data Type: Text Files
Folder: local folder
File/Table: md_regression_preference_ind_summary_closed.txt
Choose Save
Result : A regression analysis on profitable project is created and saved.
Building the Clustering Model
From the home screen of SAP Predictive Analytics, select the Modeler section
Select Create a Clustering Model
In the Select a Data Source window,
a. Choose Use a File or a Database Table option
b. Data Type: Database
c. Folder: select Browse and connect to your SAP HANA instance using the <Domain User> logon
d. Data Set: select the Browse button to select the data set that will be used for predictive
modeling.
Data manipulations appear at the top of the list of options
Choose
DM_PROJECT_TOTAL
Choose Next to continue.
Enter the Select Closed Project? prompt value to filter the data set i.e. Closed = 1
After entering the value, choose OK to begin modeling
Choose Analyze and a description for the data set appears.
Choose Next when the data description accurately describes the data
The data set description was previously completed during the data manipulation steps. You will
not need to change anything in this window
Choose Next to Selecting Variables screen
In the Selecting Variables screen, select TURNOVER, PROJECT_EBIT, PROJECT_EBIT_RATIO as
the Target Variables. Select PROJECT_ID and CUSTOMERID to Excluded Variable.
Select Next. Review Summary of Modeling Parameter to set additional setting.
Set Find the best number of clusters in this range : ( 5 ; 10)
Select Generate. Engine runs the Training the Model. Review the Model Overview.
In the Using the Model screen , select Display and review the resulting model by viewing Model
Overview, Model Graphs, Cluster Profiles, Statistical Reposts, etc.
After reviewing the model results, choose Run and select Apply Model