Xperiflow Routines
Routines are core data science building blocks within Xperiflow. They encapsulate data science capabilities into accessible, reusable functions that anyone in your organization can leverage - without needing to write code.
What is a Routine?
A Routine is a packaged data science function designed to solve a specific analytical problem. Each routine accepts inputs you provide, processes them using sophisticated algorithms, and produces meaningful outputs.
Think of routines as expert data science tools: you bring the data and business context, and the routine handles the complexity of the analysis.
Examples include:
-
Clustering customers into segments
-
Detecting anomalies in time series data
-
Allocating forecasts to granular levels
-
Training and applying machine learning models
-
Cleaning and transforming datasets
How Routines Work
Working with routines follows a simple three-step pattern:
Step | What Happens |
|---|---|
| Make your own working copy of a routine from the library |
| Provide inputs and run the routine |
| Access the outputs - data, reports, visualizations |
This pattern repeats as you iterate on your analysis or apply the routine to new data.
Key Concepts at a Glance
Understanding these concepts will help you work effectively with routines:
Concept | What It Means |
|---|---|
Routine Instance | Your personal, named workspace for a routine |
Routine Run | A single execution with specific inputs |
Routine Method | A callable capability within a routine (e.g., "fit", "predict") |
Artifacts | The outputs produced by a run (datasets, reports, charts) |
Stateful vs. Stateless | Whether a routine remembers information between runs |
Input Parameters | The configuration and data you provide to a run |
Versioning | How routine updates are tracked and managed |
Memory Capacity | Resource allocation for routine execution |
Each of these topics is covered in detail in the articles that follow.
Two Types of Routines
Routines come in two flavors, each suited to different needs:
Stateless Routines - Each run is independent. Great for transformations, aggregations, and one-time analyses.
Stateful Routines - Runs build on each other. Essential for machine learning workflows where you train a model, then apply it to new data.
→ See Stateful vs. Stateless Routines for details
Finding and Using Routines
The Routine Library is your starting point. Browse available routines via the SensibleAI Studio and https://datasensesoftware.atlassian.net/wiki/x/JoBrbQ and understand what each one does before creating an instance.
Each routine in the library includes:
-
A description of its purpose
-
The types of inputs it requires
-
The outputs it produces
-
Real-world use case examples
What's Next
The following articles dive deeper into each aspect of working with routines:
-
Stateful vs. Stateless Routines - Understanding the two paradigms
-
Routine Instance - Creating and managing your workspaces
-
Routine Run - Executing and monitoring your analyses
-
Routine Input Parameters - Configuring your runs
-
Routine Artifacts - Working with outputs
-
Routine Storage Structure - How data is organized
-
Routine Memory Capacity - Resource management
-
Routine Versioning - Tracking changes over time