Optional: Formulas & multi-metrics
Learn to apply calculated metrics to analytics dashboards in NLX's Dialog Studio
Last updated
Learn to apply calculated metrics to analytics dashboards in NLX's Dialog Studio
Last updated
Formulas can be used to convert values (seconds to milliseconds) or combine multiple filtered metrics into one. This enables you to view desired data without the need to looking at several charts and manually perform calculations.
Formulas enable you to add one or more metrics to a mathematical expression to create an entirely new metric. Metrics created through a formula are called calculated metrics. You can reference existing metrics, numbers (1, -2, 4.513) or mathematical constants such as π (pi). Parentheses () can be used to denote the order of operations within your expression.
Be sure to add one or more datasets before applying a formula to your chart.
The following lists all chart types where formulas may be used:
Bar chart: Displays formula output as a new bar
Line chart: Displays formula output as a new line
Table: Displays formula output as a new column
The following lists the basic arithmetic functions supported; x, y represent constants or metric references:
| Operation | Usage | Examples |
|---|---|---|
The following lists the mathematical functions that are supported:
To add a formula to a Bar chart, Line chart, or Table:
Select an existing Analytics dashboard or + Add new. If choosing an existing dashboard, hover beside the dashboard name to select the edit option
Select + Add panel > Hover over panel > Click the edit icon
Select + Add dataset or + Add data column; Repeat as needed
Choose + Add formula or + Add formula column > Enter a name
Formula: Enter the mathematical function. To reference an existing metric, type an open curly bracket { that displays a list of available metrics to reference
Formula unit: Choose how the calculated metric is displayed. For example, if your formula outputs a percentage, you can select the Percentage option from the dropdown to display a percent % character next to your numerical output
When done, click Save
Referencing a calculated metric into another formula input is unavailable.
Analytics can support multiple datasets within the same panel, allowing you to compare different metrics against each other in the same chart without the need to create individual charts for each metric. For example, you may want to compare the performance of different intents against each other over the same period of time.
Multi-metrics are currently only supported for Bar charts, Line charts, and Tables.
As an example, let's compare the number of actions invoked versus the number of data requests invoked in a single Line chart:
Select an existing Analytics dashboard or + Add new. If choosing an existing dashboard, hover beside the dashboard name to select the edit option
Select + Add panel > Hover over panel > Click the edit icon
Enter a name & description of the panel
Select Line chart for Chart type
Click + Add dataset link > Enter name (in this example, we'll name it Actions)
Select the metric for the data to be measured, which in this case is Actions invoked
Select the time interval of how the data will be displayed in the line chart. Day will show the results on a daily basis
Select the sum aggregation to get the total counts of actions invoked
Select respective filters to create any subset of data
Click + Add dataset > Enter name (in this example, we'll name it Data requests)
Repeat steps above as done for the Actions dataset, except select the metric Data requests invoked
When done, click Save
| Operation | Usage | Examples |
|---|---|---|
Addition
+
1+1 {SampleMetric}+1
Subtraction
-
10-8 {SampleMetric}-20
Multiplication
*
2*3 5*{SampleMetric}
Division
/
1/2 1/{SampleMetric}
Unary subtraction
-x
-1 -{SampleMetric}
Absolute value
abs(x)
abs(-1) abs({SampleMetric})
Square root value
sqrt(x)
sqrt(4) sqrt({SampleMetric})
Ceiling function
ceil(x)
ceil(1.67) ceil({SampleMetric})
Floor function
floor(x)
floor(3.2) floor({SampleMetric})
Power function
power(x,y)
power(2,2) power({SampleMetric}, 10)
