Setting clustering and centrality measures

Changing the centrality measure that is used to analyze a chart provides different information about your chart items.

Procedure

To change the centrality measure:
  1. Click the Analyze tab, and then in the Gain Insight group, click Social Network Analysis.
    The Social Network Analysis pane opens.
  2. Click the Options tab.
  3. Optional: In the Clustering section, select the K-Core checkbox to identify clusters of linked items.
  4. In the Centrality section, select the check box for the measures that you want to calculate:
    OptionDescription
    Betweenness To find gatekeeper entities that might control information flow. This checkbox is selected by default. If betweenness is selected, you can also select Betweenness for Links to find links that might be the most influential connections in the network.
    Closeness To find entities that might access the rest of the network more quickly than others.
    Degree To find the entities with the highest number of direct links to other entities in the network.
    Eigenvector To find the entities with direct links to the most influential entities in the network.
    Note:
    At least one check box must be selected for analysis to take place.
  5. Select the Use Link Directions checkbox if you want to take directed links into consideration when you calculate centrality measures. If arrows are not displayed on a link, it is assumed that information flows in both directions.
  6. Optional: Leave the Enhanced Analysis option selected to use the advanced mathematical algorithms to calculate results. If your chart contains several unconnected networks, or directed links that block paths between certain entities, Analyst's Notebook takes unconnected networks, or directed links into account and adjusts the calculations. The results reflect the relative importance of these entities and paths to those entities and links in the entire chart.
    Note:
    The only time that you clear the Enhanced Analysis option is if it is important that you do not use the mathematical algorithms that are supported in Analyst's Notebook.
    For example, to maintain a consistent set of results. No adjustment is made to take into account directed links that block paths between entities. The results that are calculated for separate unconnected networks reflect the relative importance of the entities within each network. The links are not relative to the entities on the chart.
  7. Normalize Results to adjust the calculations and display them as percentages in the Results table. If you clear this option, no adjustment is made to the calculations and results are displayed as raw data. Normalizing the results makes it easier to compare the centrality results of items in different networks and charts. Click Advanced Options and choose between two different methods of adjustment. The adjustment applies to all the selected centrality measures:
    OptionDescription
    Use the standard method Results are calculated and then divided by what, in theory, is the highest possible result for each centrality measure. They are then expressed as percentages in the Results table. This method is the standard normalization method that is most commonly used in social network analysis.
    Normalize results relative to the maximum value Results are calculated and the highest result or actual maximum for each centrality measure is determined. This value is displayed as 100% in the Results table. Every other value is divided by the actual maximum and displayed as percentages relative to 100% in the Results table.
    Note:
    The Normalize Results option is not available if you cleared the Enhanced Analysis option. Your results are calculated and adjusted by using the methods that are supported in Analyst's Notebook.
  8. Use the Actions on Completion section to specify:
    1. Whether the results are displayed on your chart by selecting Show Results on Chart
    2. Whether the appearance of chart items changes based on the analysis by selecting Apply Conditional Formatting.