Social Network Analysis include below two node for social data analysis.
The Group Analysis node imports call detail record data from a fixed-field text file, identifies groups of nodes within the network defined by the records, and generates key performance indicators for the groups and individuals in the network. See the topic Group analysis overview for more information.
The Diffusion Analysis node imports call detail record data from a fixed-field text file, propagates an effect across the network defined by the records, and generates key performance indicators summarizing the results of the effect on individual nodes. See the topic Diffusion analysis overview for more information. Processing of the IBM SPSS Modeler Social Network Analysis nodes must be done by IBM® SPSS® Modeler Server Social Network Analysis. When executing a stream containing these nodes in SPSS Modeler, you must connect to a IBM® SPSS® Modeler Server instance that also includes IBM SPSS Modeler Server Social Network Analysis.
The Group Analysis node, which is available from the Sources palette, identifies sets of individuals in a network that are socially similar to each other and determines the relative social status of the individuals in the groups. Groups that are relatively small, have many connections between members, and include a strong leader are at an increased risk of churn due to group dynamics. You can use the results to determine group churn scores that, when combined with individual churn scores, improve the ability to predict churn over models based on individual scores alone. Given the amount of data that typically comprises the network and the nature of the social interactions underlying the group definitions, group analysis is a time-consuming process. Typically, you would save the group analysis results to a database or file to be used as input to a predictive model. The group results would be refreshed relatively infrequently, such as monthly, while the model may be refreshed much more often. Requirements. The node requires a fixed width text file defining the social network using three fields. One field identifies the source for each directed relationship, one field defines the destination for each directed relationship, and the third field specifies an optional strength for each relationship. All relationships in the network must be directional.
The Diffusion Analysis node, which is available from the Sources palette, propagates an effect from a specified set of individuals across a social network, using the network relationships to identify the individuals most impacted by the effect. If the effect is churn, for instance, the node identifies those individuals most likely to churn due to other specific individuals in the network churning. You can augment existing models and data with the node output to improve the predictions generated by those models. For example, you can update individual churn propensity values to include diffusion influences. Requirements. The node requires two fixed width text files containing the data to be analyzed. The first file defines the social network using three fields. One field identifies the source for each directed relationship, one field defines the destination for each directed relationship, and the third field specifies an optional strength for each relationship. All relationships in the network must be directional. The second file contains a list of identifiers from which the effect should begin.