Creating a network dataset

Once you have developed your network dataset design, you are ready to create a network dataset.


There are two exercises in the Network Analyst tutorial that step through creating a network dataset. You can link to them from the Related Topics section of this document.

There are five main steps to creating a network dataset:

  1. Prepare the feature dataset and sources.
    • If you are creating a geodatabase-based network dataset, all feature classes participating as sources in a network should be present in one feature dataset.
    • If you're creating a network from shapefiles, you need to organize all the feature classes that will participate as sources into the same folder.
    • You may need to create new feature datasets or new feature classes, and you may need to edit features.

  2. Prepare the sources for appropriate roles inside the network dataset.
    • Be sure your sources have fields that represent your network impedance values—distance, travel time, and so on. For best results, name these fields as the units of your impedances, since these fields will automatically be detected by the New Network Dataset wizard—for example, you might want to name your travel time field Minutes. For edge sources, if the impedance values differ based on direction of travel, provide a separate field for each direction of travel, for example, FT_Minutes and TF_Minutes.
    • If you are modeling one-way streets, be sure that your edge sources have a field specifying one-way street information. The New Network Dataset wizard will recognize a string field named One_Way or Oneway, creating evaluators that interpret its values as follows:

      • FT or F indicates a one-way street permitting travel only in the digitized direction of the edge.
      • TF or T indicates a one-way street permitting travel only against the digitized direction of the edge.
      • N indicates a street that does not permit travel in either direction.
      • Any other value indicates a street that permits travel in either direction.

      Learn about types of evaluators used by a network

    • If your data uses z-elevation or z-level values to model overpasses and underpasses, be sure this information is stored in a pair of integer fields—one field for each end of the edge. If the fields are named either F_ELEV and T_ELEV or F_ZLEV and T_ZLEV, the New Network Dataset wizard will automatically detect these fields.

      Learn about connectivity

    • If you plan to generate driving directions text from your network analyses, be sure that your edge sources have fields containing the information needed to generate driving directions such as road class, street names, highway shields, and boundary information.

      Learn about setting directions

  3. Prepare turn feature classes and add turn information.
    • The turn feature class is associated with a network dataset.
    • Turns are not necessary to create a network dataset.
    • If your turn information is stored in an ARC/INFO or ArcView GIS turn table, import the turn table to a turn feature class.

      Learn more about migrating turn data

    • If you do not have ARC/INFO or ArcView GIS turn tables but want to use turn information for network analysis, create a new turn feature class and add new turn features to store that information.

    • Be sure that your turn feature class contains fields with information that will be used in the network attributes, such as turn impedances (for example, this turn takes two minutes to make) or conditional turning restrictions (for example, only trucks cannot make this turn).
  4. Create the network dataset using the New Network Dataset wizard.
    • The New Network Dataset wizard walks you through naming the network dataset, identifying the network sources, setting up the connectivity, identifying elevation data (if necessary), specifying turn sources (if necessary), defining attributes (such as costs, descriptors, restrictions, and hierarchy), and setting up the direction-reporting specifications.
  5. Build the network dataset.

Related Topics