## Causality

Causality goes deeper than just correlations. It establishes relationships of cause and effect in the data into cause and effect, which helps us identify the root cause of real world problems.

ROOT CAUSE ANALYSIS

Automatically identify the real world root cause of an anomaly in your data

Causality is a method of statistical inference designed to find root causes of events and anomalies.

Given that finding these causes is a complex process which is growing exponentially with the number of possible causes, we follow a 2-step approach which first constrains the possible causes to the most significant ones (→ “Correlation spotting”) and then determines the cause using a Bayesian network involving a causal graph and conditional probabilities (→ “Causal reasoning”).

APPROACH

Our 2-step approach to root cause analysis

#### 1. Correlation spotting

The possible cause(s) of an anomaly can be based on one or more dimensions, such as installed software version, recent OS upgrade, usage load, or location, to name a few. The complexity grows exponentially with the number of causal dimensions. Therefore, in a first step correlation spotting is applied to establish a correlation between anomalies and most impactful causal dimensions.

#### 2. Causal reasoning

We apply causal reasoning to ensure you confidently take the actions that will have the greatest impact. Making a wrong decision, may lead to the wrong approach for how to solve a problem. The causality algorithm first creates a causal graph based on the statistical history of the causal dimensions. In a second step conditional probabilities are calculated and analyzed. The cause with the highest probability is considered to be the root cause for the anomaly.