Most functions in tsauditor/profiler/, tsauditor/anomaly/, and tsauditor/leakage/ expect a cleanly structured tabular DataFrame. However, many real-world time series metrics start as unstructured text logs (e.g., application crashes, server syslogs).
What’s needed: Create a new sub-module tsauditor/parsers/ that handles unstructured text ingestion. This module will read raw log streams, parse out timestamps and categorical variables using regular expressions, aggregate occurrences into standard time frequencies, and return a fresh tabular DataFrame.
Why this is approachable: This keeps our core engine completely isolated. Instead of forcing our core analytics or remediation code to handle unstructured strings, this feature serves as a clean, non-destructive gateway that shapes messy logs into tabular frames before an audit ever starts.
Most functions in tsauditor/profiler/, tsauditor/anomaly/, and tsauditor/leakage/ expect a cleanly structured tabular DataFrame. However, many real-world time series metrics start as unstructured text logs (e.g., application crashes, server syslogs).
What’s needed: Create a new sub-module tsauditor/parsers/ that handles unstructured text ingestion. This module will read raw log streams, parse out timestamps and categorical variables using regular expressions, aggregate occurrences into standard time frequencies, and return a fresh tabular DataFrame.
Why this is approachable: This keeps our core engine completely isolated. Instead of forcing our core analytics or remediation code to handle unstructured strings, this feature serves as a clean, non-destructive gateway that shapes messy logs into tabular frames before an audit ever starts.