Correlations

About the narrative

Details covered in the analysis

The Correlations app detects correlations in a dataset and generates a narrative that describes them in terms of strength (very weak to very strong) and direction (positive or negative).

A positive correlation is indicated when the values in two data series are both high, both low, or are changing in the same direction. For example, Profit may be high when Sales are high.

A negative (or inverse) correlation is indicated when the values in one data series are high/rising as the values in the other data series are low/falling (and vice versa). For example, Profit may decrease as Cost of Goods Sold increases.

 

The app can be used to analyze:

One measure, to identify correlations between different dimensional breakdowns. For example:

  • For non-time series data:

    arria-apps-correlations-one-measure.png
  • For time series data:

    arria-apps-correlations-one-measure-time.png

Two measures, to determine if they are correlated in terms of overall values and across different dimensional breakdowns. For example:

  • For non-time series data:

    arria-apps-correlations-two-measures.png
  • For time series data:

    arria-apps-correlations-two-measures-time.png
  • For time series data with time shift:

    arria-apps-correlations-two-measures-time-shift.png

Important

A correlation between measures does not necessarily mean that a change in one measure is caused by a change in the other.

 

Top

Data requirements

All NLG Apps have the same data requirements for supported formats and field names. See NLG Apps Data requirements for details.

In addition, each app has specific data requirements:

If your data selection and configuration do not meet the specific requirements for a particular app, the app is disabled in the wizard and a warning message is displayed:

arria-apps-disabled-correlations.png

Dimensions and measures

The number of dimensions and measures required depends on whether you're analyzing time series data or non-time series data.

Tip

When looking for correlations between two measures, set one as the Primary Measure and one as the Secondary Measure in Step 2 of the wizard to determine the order in which they are compared. The selected order is particularly important when considering whether a change to one measure may be affecting the other.

 

Non-time series data

If your data selection does not contain a time dimension, this app can analyze:

  • up to two measures and up to three non-time dimensions.

Minimum requirements:

  • one measure and two non-time dimensions OR

  • two measures and one non-time dimension.

Important

At least four data points are required in each series. If insufficient data is selected, an error is generated.

A maximum of 20 instances of each dimension is reported in the narrative.

Time series data

If your data selection contains a time dimension, this app can analyze:

  • up to two measures, up to three non-time dimensions, and multiple time dimensions*.

Minimum requirements:

  • one measure, one non-time dimension, and one time dimension* OR

  • two measures and one time dimension*.

 

*Time dimension requirements for each analysis period:

Analysis period

Required date fields

Year

Date or Year

Quarter

Date or Year+Quarter

Month

Date or Year+Month

Day

Date or Year+Month+Day

 

Important

At least four data points are required in each series. In addition, at least two instances of each non-time dimension are needed.

A maximum of 20 instances of each dimension is reported in the narrative.

Aggregation and entity types

Not all combinations of aggregation and entity types are supported for measures. The Correlations app can analyze the following combinations:

arria-apps-entity-aggregation-correlations-ms.png

Note

Entity types and other data attributes are set in Step 2 of the wizard.

 

Top

Narrative options

The following fields are available to configure in Step 3 of the NLG Apps wizard: Narrative options.

OPTION

DESCRIPTION

LOCATION

Period

Select a time period to analyze. The time dimensions in your data selection must include the period you select here — either as a Date field, or a Day, Month, Quarter, or Year field.

Default: Month

tableau-arria-period.png

Time Shift Interval

When looking for correlations between two time series, select the interval (number of periods) by which to shift the secondary time series. If a correlation is found, it may indicate that the primary series has a delayed effect on the secondary series.

A time shift can only be applied to correlation analysis if your data contains N consecutive data points from the start of the series, where N is the number you select in this field.

This field is available only when both a primary and secondary measure have been configured in Step 2 of the wizard, and your data selection contains a time dimension (time series data).

Default: 0

arria-apps-options-time-shift-interval.png

 

Top