Flow Relational Model

Overview #

A Flow Relational Model is a specialized transform that produces a data structure which can be visualized using a Flow Diagram. The important concept behind a Flow Relational Model is that it builds up a hierarchical dataset based off of link information. This differs from a Tree Map Relational model, which builds its model from node information.

Create a Flow Relational Model Transform #

  • Select the  icon for a node,
  • Select the “Transform” option from the context menu.

The “Choose a Transform” dialog will appear, showing the available transforms in the system.

  • Select the “Flow Relational Model” option. This will launch the Flow Relational Model Wizard:

Flow Relational Model Wizard #

Step 1:  Configure #

Property
Description
Name A symbolic name used to represent this transform. This is the name that will show up the in the pipeline node representing this transform.
Description This is where the administrator can enter notes for the transform.

Step 2:  Columns #

This step sets up the relationship between attributes in the underlying dataset.

Property
Description
Column This list of attributes defines the relationship to be displayed inside of corresponding Sankey and Chord Diagrams. These attributes defined here will show up as:

  • Columns within the Sankey visualization
  • Separate Groups on the outside of a chord diagram.

The unique values for each attribute will show up as:

  • Nodes within a the individual columns inside a Sankey visualization
  • Nodes on around the outer perimeter of a Chord diagram.

If you were looking as Stock Market data (see example below), you may want to see Summations of stock prices by “Sector”, and how those sectors feed “Industry”, and ultimately how the different Industries affect the various “Markets”. The order can be adjusted by dragging list items to new positions within the list. The size of the nodes will be calculated based on the Link Value Aggregation method selected. See “Link Value Aggregation” below.

Link Value Aggregation

This property determines how to calculate the proportional sizes of the nodes inside of a Sankey or a Chord Diagram. The visualizations for a Flow Relational Model will group the data by each specified attribute, and then generate a proportional size based on the aggregation function specified here. There are several different options for calculating sizes:

  1. Count
    This determines the size of nodes based on the total number of records for each unique value in the underlying data. Assume Stock Data (see example below), and that node sizes should be based on the total number of listings for each “Market”. A “COUNT” aggregation will provide this type of calculation. If this was real stock data, then it would result in the following:

    1. The “NYSE” value from the “Market” attribute would have ~2,800 records, one of each company listed.
    2. The “NASDAQ” value from the “Market” attribute would have ~2,900 records, one for each company listed.
  2. Sum
    This determines the size of nodes based on the summation of values for an attribute specified in the underlying data. When using SUM, you must select an attribute that has a NUMBER value. Assume Stock Data (see example below), and that node size based should be based on the on the price per share for all listings per market, rather than the total number of listings. To do this, select “SUM”, and then select the “Stock Price” column. If this was real stock data, then it would result in the following:
    1. NYSE: Sum of all stock prices at that time.
    2. NASDAQ: Sum of all stock prices at that time.
  3. Max
    This determines the size of nodes based on the maximum value for an attribute specified in the underlying data. Assume Stock Data again (see example below), but this time node size should be based on the MAX Stock Price per market. If this was real stock data, you would see:

    1. NYSE: $249,270.00
    2. NASDAQ: $807.22
  4. Min
    This determines the size of nodes based on the minimum value for an attribute specified in the underlying data. A MIN calculation is not very useful when looking at Stock Markets, but if you were looking at interest rates across banks, a MIN value calculation might be useful.
  5. First
    This determines the size of nodes based on the value in the first record. This is useful when a single value is repeated for all records.

Sample Stock Data #

Name
Market
Sector
Industry
Price
APPL Apple Inc. Nasdaq Information Technology Technology Hardware & Equipment 113.00
AMZN Amazon.com, Inc. Nasdaq Consumer Discretionary Retailing 759.62
GOOG Alphabet Inc Class A Nasdaq Information Technology Software & Services 807.22
MCD McDonald’s Corporation Nasdaq Consumer Discretionary Consumer Services 121.57
BRK/A Berkshire Hathaway Inc. NYSE Financials Diversified Financials 249,290.00
TWX Time Warner Inc. NYSE Consumer Discretionary Media 94.71
VZ Verizon Communications Inc. NYSE Telecommunication Services Telecommunication Services 51.90
PEP PepsiCo Inc. NYSE Consumer Staples Food Beverage & Tobacco 104.31

Step 3:  Upstream Variables #

Details about this step can found here: Satisfy Upstream Node Variables Wizard Step

Step 4:  Preview #

Refer to the “Variables” section for more information.