02 Transformations

Transform and process data in Grafana.

Transformations process data after the query but before visualization: aggregate, filter, rename, combine. In this tutorial you will learn the most important transformations and build a summary table like this step by step:

What you will learn

• Aggregate data with Reduce (minimum, maximum, mean)

• Rename, reorder, and hide fields

• Add trend lines with regression analysis

• Combine data from multiple queries

• Chain transformations

Prerequisites

• Completed Part 1 and Part 2 of the tutorials

• Basic knowledge of panel creation

---

The Transform tab

1. Edit a panel

2. Click the Transform tab (between Query and Alert)

3. Click + Add transformation and select a transformation

Transformations are executed from top to bottom. You can reorder them via drag and drop, disable them (eye icon), or delete them (trash icon).

The order affects the result. A filter before Reduce produces different results than after.

---

Reduce

Converts time series data into aggregated values. Ideal for summary tables.

1. Add the Reduce transformation

2. Select the calculations:

Calculation	Description
Last *	Last non-null value
Min / Max	Minimum / Maximum
Mean	Average
Sum	Sum of all values
Difference percent	Percentage change
All values	Array of all values (for sparklines)

---

Organize fields by name

Rename, reorder, and hide columns. Essential for clean tables.

1. Add the Organize fields by name transformation

2. For each field:

   • Rename: Enter a new display name

   • Hide: Click the eye icon to hide

   • Reorder: Drag and drop to reorder

---

Rename by regex

Rename multiple fields at once using pattern matching.

Setting	Description
Match	Regex pattern to match
Replace	Replacement text ($1, $2 for capture groups)

Examples:

Match	Replace	Before	After
Extruder (.) Durchsatz	Ext $1	Extruder A Durchsatz	Ext A
\(Soll\)	(Target)	Speed (Soll)	Speed (Target)
(.+) - (.+)	$2	Kiefel - Durchsatz	Durchsatz

Common regex patterns

Pattern	Description	Example
.	Any single character	A, 1, -
.*	Any number of characters	Extruder A Durchsatz
.+	At least one character	like .*, but not empty
(...)	Capture group (use $1, $2 in Replace)	(Extruder .) → $1
\( / \)	Literal parentheses	\(Soll\) matches (Soll)
^ / $	Start / end of string	^Kiefel matches only at the start
[A-C]	Character class	Extruder [A-C]
\d+	One or more digits	Sensor \d+ matches Sensor 42

---

Filter by name

Show only specific fields in the output.

1. Add the Filter by name transformation

2. Select Include (show only selected fields) or Exclude (hide selected fields)

3. Select fields manually via checkboxes or filter by regex

Regex examples for Filter by name:

Mode	Regex	Effect
Include	Durchsatz|Temperatur	Show only throughput and temperature fields
Exclude	.*\(Soll\)	Hide all setpoint values
Include	Extruder [AB].*	Show only Extruder A and B

With Filter data by values you can also filter rows by conditions, for example only showing data points where the temperature is above 200 °C.

---

Regression analysis

Add trend lines to identify how data evolves over time.

1. Add the Regression analysis transformation

2. Set X field to the time field and Y field to the value field

The regression line shows the overall trend: rising, falling, or stable.

Style the trend line

By default, the regression line has no unit and appears on a separate Y-axis. Add an override:

1. Match the field by regex: .*predicted.*

2. Set the Unit to the same unit as the original data

3. Choose a different color (for example red) and a dashed line style

---

Combine data

When you want to merge data from multiple queries, two transformations are available:

Transformation	Description
Merge	Combines all time series into one table. No configuration needed.
Join by field	Joins tables on a shared field (for example a timestamp). Mode: Inner, Left outer, or Right outer.

---

Example: machine comparison

This example shows how to use a transformation chain to summarize the throughput of two machines in a compact comparison table.

Step 1: Create two queries

Create two separate queries, one for each machine. Name the queries after the machine (for example "Kiefel" and "Macchi") so the data can be distinguished later.

Step 2: Build the transformation chain

Switch to the Transform tab and add the following five transformations one after another:

1. Rename by regex (x2): Rename the "Durchsatz" field in each query so the machine is identifiable. Match: Durchsatz, Replace: Kiefel or Macchi. Click the filter icon on the transformation and under Apply transformation to select the corresponding query (for example "Query: Kiefel"). This way each rename only applies to its own query.

2. Join by field: Joins the two renamed time series on the timestamp into a single table.

3. Reduce: Aggregates the time series into one row per machine with the calculations Last *, Min, Max, Mean, All values, and Difference.

4. Organize fields by name: Renames the columns to be more readable (for example "Field" to "Machine", "Last *" to "Current", "All values" to "Trend") and sorts them into the desired order.

Result

The finished table compares both machines at a glance, including a sparkline trend display:

The column formatting (Bar Gauge for "Current", colored background for "Change", Sparkline for "Trend") is configured via Overrides > Fields with name > Cell options. For details, see the tutorial Tables.

Use the Table view toggle in the panel editor to check intermediate results after each transformation.

---

Tips

• Performance: Transformations run in the browser. Filter at the query level when possible.

• Debugging: Add transformations one at a time and use the Table view toggle to check intermediate results.

• Common patterns: Time series to summary table: Reduce + Organize fields. Sparklines: Reduce with All values + Override on cell type.

---

Next steps

• Introduction to the ENLYZE API - API-based dashboards for advanced use cases