MyObservability

Predict Analytics

Overview

Predictive Analytics uses machine learning algorithms to predict the health score value of a selected service. The models use historical service health score and KPI data to approximate what a service’s health might look like in 30 minutes.

Set up Predictive Analytics

You need to coinstall Splunk Machine Learning Toolkit (SMLT) & install Python for Scientific Computing add-on, Optionally set up permissions.

Permissions:

• itoa_admin: have access to train, test, add, create alerts & delete model.
• itoa_team_admin: have access to train, test, add, create & delete model to the service owned by team.
• itoa_analyst: have access to read, test, create alerts & add model to glass tables.
• itoa_user: Read, test & create alerts from model.

Checking if service is fit for PA

ITSI provides two visualizations to help you decide whether your service is a good fit for predictive modeling.

• Service Health Score and KPIs over time: Graph displays the values of service health scores and KPIs over a selected time period. If you see unusual values or outliers in your data, confirm whether these data points are relevant and real.
• Distribution of Service Health Score Values: histogram of service health score values over a selected time period.

Train a Predective Model

• Specify a time period.
• Choose an algorithm type.
  • Regression Algorithems
    • Linear Algorithem
    • Random Forest Algorithem
    • Gradient boosting Algorithem
    • Logistic Algorithem
  • Classification Algorithems
• Choose a machine learning algorithm.
• Split your data into training and test sets.
  • Training set (Default 70%)
  • Test set (Default 30%)

Test a Predective Model Always evaluate a model to determine if it will do a good job predicting future health scores.

It is advised to only save one service at a time when creating models. Saving multiple models in separate windows might cause an error.

Types

• R2
• RMSE

---

Next Chapter: Glass Tables

---