Network Extractor enables you to quickly and easily generate Network Graphs from tabular datasets. A Network Graph is an incredibly powerful tool for visualizing similarities and identifying groups, or communities, within your data. 

Input a target data feature into Network Extractor, and that data feature will become the nodes in the network. Whichever associative data features you choose will then help to define the edges (similarities) in the network. Once a Network Graph is generated, you can utilize Network Analysis Tools to understand the characteristics of the communities within your data.

• Related Article: Reviewing Network Data with Network Analysis Tools and Network Insights

Important: We recommend running Network Extractor on datasets with fewer than 20,000 - 30,000 nodes for fast runtimes.

Using Network Extractor

1. Click the Network Extractor icon ( ) in the top toolbar to open the Network Extractor panel.

2. Click and drag a feature from the Features Panel to the input box below Add Target to define your nodes.

   • The Target can be any feature in your dataset that can be used to distinguish between nodes in your network. Often, some type of unique identifier (ID) field works best.

   • This field can be unique or non-unique. If there are duplicate values in the Target feature, the dataset will be aggregated using the method selected in the second dropdown list (this defaults to Mean).

3. In the Associative Features section, choose whether you'd like to use a Categorial or Numerical feature set to generate your Network Graph.

   • These features will be utilized to construct associations between the nodes, signifying similarity. 

4. Add other features by choosing them from the Select Features drop-down or using the Input All button.

5. (Optional) Remove unwanted features by hovering over the feature and clicking the Remove ( ) button. 

   Tip: Any rows with missing values will not be visualized in the Network Graph, so you may want to remove sparse features (with many missing values). Click the red pound sign ( ) to remove sparse features.

6. Once the Target and Associative Features are determined, click Run at the bottom of the panel.