In a world dominated by tabular data, network data offers an alternative way to view and analyze connected data points. However, working with network data presents its own set of challenges: how to scale, make it actionable, and even how to get started. In this guide, we’ll explore how Knowi simplifies the complexities of network data analysis and take you through the steps of building a telecommunications network from the ground up.
What is Network Data?
Network data refers to structured data that represents relationships between entities, typically visualized as graphs consisting of nodes (vertices) and edges (links). Each node represents an entity, such as a person, device, or organization, while edges define the relationships or connections between them.
Real-World Examples:
- Social Networks: Platforms like Facebook or LinkedIn use network data to represent how users are connected through friendships, follows, or professional relationships.
- Telecommunication Networks: Nodes are hardware or endpoint devices (servers, routers, phones), while edges are the connections (fiber, internet, SD-WAN links) that facilitate communication between them.
- Supply Chains: Nodes represent suppliers, manufacturers, and distributors, while edges illustrate the movement of goods through the chain.
Network analysis helps businesses and organizations discover key influencers, trace connections, and optimize paths between nodes. It can also uncover hidden patterns, such as clusters of related entities, or even highlight vulnerabilities in systems like security networks.
The Challenges of Working with Network Data
Network data provides valuable insights that are difficult to obtain through other means, but managing it efficiently comes with its own set of challenges.
Challenge: Getting Started
Building graph networks can be challenging for non-technical users due to the need for properly formatted data, installation and setup difficulties, and the complexity of graph theory concepts. Many platforms require coding knowledge or the ability to customize layouts and configurations, which can be overwhelming for those without programming experience.
Solution: Easily Connect with Knowi
If you’re working with a relatively simple dataset and just want to get started quickly, you can easily upload a CSV and visualize your network right away. For more complex, enterprise-level challenges, Knowi integrates with TigerGraph, enabling powerful analysis for larger datasets.
Challenge: Scalability
An inherent issue that arises with the interconnectedness of graphs, is that they can grow to a scale that is no longer useful, making storage, querying, and visualization more complex. Large-scale networks with thousands or millions of nodes and edges can become overwhelming without a way to break these networks down into more manageable chunks.
Solution: Filtering the Network
With Knowi, you can apply dashboard-level filters to display specific segments of the network. If you include a filterable categorical field in the network dataset, you can narrow the view to show only a portion of the network, improving readability.
Challenge: Lack of Interactivity
Traditional network visualizations show entity connections and basic attributes but often lack deeper insights into relationships. These visualizations are typically static, limiting interaction with the underlying data, which makes it difficult to draw out actionable insights from the connections.
Solution: Linking the Graph Network to Additional Data
Knowi enables interactivity in graph networks with its drilldown feature, allowing users to explore network data beyond the initial visualization. By clicking on a node or edge, users can be taken to a new visualization of related data. For example, in a social network, clicking on a connection between two users can show the history of interactions between them, or in a supply chain network, clicking on an edge can reveal specific transaction details between suppliers.
Case Study: Visualizing A Telecommunications Network
This walkthrough will guide you through the process of formatting data, connecting the data to Knowi, and visualizing a graph network. We’ll cover how to use graph-specific features to customize and filter your network for clearer insights.
Step 1: Formatting the Data
To visualize your graph, you first need to format your data into a tabular structure, organizing it into vertices (nodes) and edges (links). Here’s how you can format the table:
- Vertices represent the points or objects in your graph.
- ResultName: Label this as “VertexResult”.
- v_id: The unique identifier for each vertex (String).
- v_type: Defines the type of each vertex (e.g., “Customer,” “Device”).
- attributes: A string or JSON field with additional data about the vertex, which can be displayed as a tooltip in the graph.
- Edges represent the relationships or links between vertices.
- ResultName: Label this as “@@edges”.
- from_id: The source vertex ID where the link starts.
- from_type: The type of the source vertex.
- to_id: The destination vertex ID where the link ends.
- to_type: The type of the destination vertex.
- e_type: The type of relationship (e.g., “ConnectedTo”).
- attributes: A string or JSON field with additional edge data, also passed as a tooltip.
You can store both vertices and edges in a single file, or in separate files that are later joined. The ResultName and attributes columns will contain data from both node and edge rows. For more details about the graph widget type, check out our documentation.
Step 2: Connecting to the Datasource
Once your data is formatted, you can upload it to the platform:
- Navigate to Queries → New Datasource → Upload file (or paste data).
- Alternatively, you can connect directly to a TigerGraph instance to source your data. For details on how to connect to TigerGraph see our documentation here.
Step 3: Customizing Your Graph Visualization
With the data loaded, you can start creating a graph widget to visualize your network. Follow these steps:
- Create the Widget: Use the dataset you uploaded and set the widget type to “Graph.”
- Set Edge Types: This option lets you filter the displayed edges by type. If no edge type is specified, all links will be shown.
- Integration Type: Decide how vertices are positioned in the graph. Choose between:
- Verlet (default): Applies inertia by using both previous and current positions.
- Euler: Stores and applies velocity to calculate new positions.
- Pick Custom Colors: Customize the colors of the nodes for a visually appealing and intuitive view of your graph.
Step 4: Filtering the Data (optional)
For larger or more complex graphs, viewing everything at once can be overwhelming. To manage this, you can add a categorical column to your dataset (the same csv file outlined above) to break the graph into smaller groups. For example, if you’re working with customer data, you might add a CustomerID column.
- Be sure to populate this column for all vertexes and edges you would like included in the filtered view.
- Set a filter at the dashboard or widget level for this category. (See image below for how to set up the dashboard level filter)
- When a specific CustomerID is selected, only the relevant network will be displayed, helping you focus on specific segments.
Step 5: Adding Interactivity (optional)
Graph networks provide an overview of relationships, but sometimes you need more context. With the drilldown feature, you can explore deeper insights by selecting specific parts of the graph. In the example below, a drilldown is set up so that when a node in the network is selected, a linked visualization displays the broadband speed of that node over the past month. This interaction allows you to dynamically explore relationships within the graph while connecting the data to other meaningful visualizations.
By clicking on the “SDWAN” vertex, it triggers another visualization—a line chart showing the internet speed over time specifically for that SDWAN.
A short video showcasing a drilldown from a graph network to a line chart.
Conclusion
The strategic utilization of network data can be a powerful tool for businesses seeking to understand complex relationships within their data. If you’d like to explore this further, try walking through the case study example yourself with a 21-day free trial of Knowi. And if you need any assistance, our support team is always here to help.
What is Knowi? Knowi is a business intelligence (BI) platform based in Oakland, California. Unlike other BI tools that focus primarily on visuals, Knowi takes a data-first approach, allowing it to seamlessly connect to any type of data, from any source. This makes it particularly well-suited for complex use cases such as graph networks.
