Optimize ServiceNow ML Training for ServiceNow System Admins new to Predictive Intelligence the glide.platform_ml.api.csv_min_line system property in ServiceNow Machine Learning (ML) capabilities determines the minimum dataset size required for model training. By default, ServiceNow sets this threshold to 100,000 rows, potentially delaying model training and AI-driven decision-making.

You are thrilled to begin your Machine Learning training! You dive right in with the out-of-the-box training examples, eager to explore the Predictive Intelligence Homepage. Starting with Classification, you leverage machine-learning algorithms to automatically set field values during record creation—for example, assigning an incident category based on the short description. As you train predictive models, they become intelligent agents, dynamically categorizing and routing work based on your past record-handling experience.

Clicking into Classification, you eagerly run a few of the out-of-the-box machine learning solutions, excited to see them in action. As you launch your first model, anticipation builds—only to be met with an error message. It turns out there aren’t enough records to proceed. While unexpected, this moment offers a valuable learning opportunity, pushing you to explore data requirements and refine your approach before moving forward.

What can I run in classification, that requires less records as my instance is new?

Why Does This Matter?

• Speed and Efficiency: Lack of large datasets slows time to implement, increase processing time, slowing down model development.
• Resource Utilization: Training smaller datasets reduces server strain and improves responsiveness.
• Accessibility: Organizations with limited data can now train models efficiently without reaching an arbitrary threshold.

Reducing the default setting enables faster, more flexible model training. Let’s explore how to modify this system property to streamline ML workflows.

---

Step-by-Step Guide to Adjusting glide.platform_ml.api.csv_min_line

1. Navigate to System Properties

1. Search for sys_properties.list in the filter navigator and press Enter.

2. Locate or Create the System Property

• If the property exists:
  • Click on glide.platform_ml.api.csv_min_line and modify the value to a lower number, such as 5,000 or 10,000.
• If the property does not exist:
  1. Click New.
  2. Set Name to glide.platform_ml.api.csv_min_line.
  3. Set Type to integer.
  4. Set Value to a number lower than 100,000 (e.g., 5,000 or 10,000).
  5. Click Save.

3. Restart the Instance (If Necessary)

Some system properties require a restart for changes to take effect. If adjustments aren’t reflected immediately, try these:

• System Diagnostics > Cache Flush to refresh configurations.
• Run cache.do to clear outdated data.

4. Validate the Changes

• Initiate model training and verify performance.
• Check System Logs > All for errors related to ML training.

---

Conclusion: Unlock Faster AI Insights

Lowering the minimum row threshold in ServiceNow ML significantly enhances training speed, accessibility, and efficiency. By adjusting glide.platform_ml.api.csv_min_line, teams can accelerate AI model development, optimize system resources, and unlock real-time insights without unnecessary data constraints.

Take control of your ServiceNow AI training today and streamline ML workflows with a simple configuration update!

Need Help? Let us know if you run into any issues—we’re here to support your ML optimization journey!

Other Optimize ServiceNow ML Training Resources

• d