Skip to content
DataMiner DoJo

More results...

Generic selectors
Exact matches only
Search in title
Search in content
Post Type Selectors
Search in posts
Search in pages
Search in posts
Search in pages
Log in
Menu
  • Blog
  • Questions
  • Learning
    • E-learning Courses
    • Open Classroom Training
    • Certification
      • DataMiner Fundamentals
      • DataMiner Configurator
      • DataMiner Automation
      • Scripts & Connectors Developer: HTTP Basics
      • Scripts & Connectors Developer: SNMP Basics
      • Visual Overview – Level 1
      • Verify a certificate
    • Tutorials
    • Video Library
    • Books We Like
    • >> Go to DataMiner Docs
  • Expert Center
    • Solutions & Use Cases
      • Solutions
      • Use Case Library
    • Markets & Industries
      • Media production
      • Government & defense
      • Content distribution
      • Service providers
      • Partners
      • OSS/BSS
    • DataMiner Insights
      • Security
      • Integration Studio
      • System Architecture
      • DataMiner Releases & Updates
      • DataMiner Apps
    • Agile
      • Agile Webspace
      • Everything Agile
        • The Agile Manifesto
        • Best Practices
        • Retro Recipes
      • Methodologies
        • The Scrum Framework
        • Kanban
        • Extreme Programming
      • Roles
        • The Product Owner
        • The Agile Coach
        • The Quality & UX Coach (QX)
    • DataMiner DevOps Professional Program
  • Downloads
  • More
    • Feature Suggestions
    • Climb the leaderboard!
    • Swag Shop
    • Contact
      • General Inquiries
      • DataMiner DevOps Support
      • Commercial Requests
    • Global Feedback Survey
  • PARTNERS
    • All Partners
    • Technology Partners
    • Strategic Partner Program
    • Deal Registration
  • >> Go to dataminer.services

AI – What’s the minimum time frame of data to have AI capabilities?

Solved1.89K views8th March 2022AI artificial inteligence
1
Bruno Dias [SLC] [DevOps Member]2.36K 7th March 2022 0 Comments

What's the minimum time frame of data that requires to be saved in the database to benefit from the AI capabilities?

Bruno Dias [SLC] [DevOps Member] Selected answer as best 8th March 2022

1 Answer

  • Active
  • Voted
  • Newest
  • Oldest
5
Dennis Dreesen [SLC] [DevOps Advocate]496 Posted 8th March 2022 0 Comments

Dear Bruno,

Thanks for your question. Let me go over the main features to give you an idea of how much time is required for them:

Alarm focus: In principle no alarm history is required, but it will need approximately one week of learning before it will give good results. If a week of alarm history is available the first time it is activated, it will use this week for its initial learning

Alarm grouping: the current version uses metadata in the system to group together focussed alarms. It is up and running as soon as alarm focus is up and running.

Anomaly detection: You should have at least one full week of data before expecting any suggestions / anomalies in your alarm console, but the trend graph will start showing behavioral changes earlier on (the most obvious ones after 4 hours, other more subtle ones will pop up over time)

Forecasting: Forecasting is done on several aggregated levels: the real time level, the 5' averages, the hourly averages and daily averages. A forecast will be displayed on the trend graph for each level. More long term predictions (ie further into the future) will use averaged data, whether short term predictions use real time data. In order to predict, we need at least 50 points, so after 50 days you will start seeing the long term prediction in cube, after 50 hours the mid term,... The more data we have, the longer our predictions.

Proactive cap detection: this algorithm relies on the above mentioned forecasting algorithm to proactively detect issues. This means that the longer ahead you want to be able to predict issues, the more historical data you need. As a rule of thumb, if you want to start predicting issues n hours in advance, you need 10*n hours of history. In particular, you'll need about 10 days of data if you want to predict an issue 1 day ahead of time.

State icons: State icons reflect the behavior of a parameter in a recent time window. The duration of this time window is configurable, but it is set to 1 hour by default. In this case, the latest you should see state icons is after 1 hour, but typically you will see them earlier on due to a data prefetching routine.

Pattern matching: pattern matching works as soon as you have enough data to select your pattern.

Bruno Dias [SLC] [DevOps Member] Selected answer as best 8th March 2022
Please login to be able to comment or post an answer.

My DevOps rank

DevOps Members get more insights on their profile page.

My user earnings

0 Dojo credits

Spend your credits in our swag shop.

0 Reputation points

Boost your reputation, climb the leaderboard.

Promo banner DataMiner DevOps Professiona Program
DataMiner Integration Studio (DIS)
Empower Katas

Recent questions

Correlation Engine: “Test rule” doesn’t result in a hit, despite functional rule 1 Answer | 3 Votes
When using the Setter = true attribute, will the copy action always be executed first? 1 Answer | 3 Votes
Multiple Set on Table parameters for DVE’s 1 Answer | 2 Votes

Question Tags

adl2099 (115) alarm (62) Alarm Console (82) alarms (100) alarm template (83) Automation (223) automation scipt (111) Automation script (167) backup (71) Cassandra (180) Connector (109) Correlation (69) Correlation rule (52) Cube (150) Dashboard (194) Dashboards (188) database (83) DataMiner Cube (57) DIS (81) DMS (71) DOM (140) driver (65) DVE (56) Elastic (83) Elasticsearch (115) elements (80) Failover (104) GQI (159) HTTP (76) IDP (74) LCA (152) low code app (166) low code apps (93) lowcodeapps (75) MySQL (53) protocol (203) QAction (83) security (88) SNMP (86) SRM (337) table (54) trending (87) upgrade (62) Visio (539) Visual Overview (345)
Privacy Policy • Terms & Conditions • Contact

© 2025 Skyline Communications. All rights reserved.

DOJO Q&A widget

Can't find what you need?

? Explore the Q&A DataMiner Docs

[ Placeholder content for popup link ] WordPress Download Manager - Best Download Management Plugin