Description
It seems like the output dataframe of anomaly detection of timeGen has fewer rows than its input dataframe. I first made this observation in my test data and upon checking tutorial page for anomaly detection and using the example dataframe in this page, I observed the same thing. For example, df in the anomaly detection tutorial page starts at ds = 2007-12-10, while anomalies_df starts at ds = 2008-01-10. However, nothing is mentioned in the documentation. It would be very helpful to add how many data points will be missing from the anomaly detection output (I guess there should be a formula for that) and what is the logic behind it.
Link
No response
Description
It seems like the output dataframe of anomaly detection of timeGen has fewer rows than its input dataframe. I first made this observation in my test data and upon checking tutorial page for anomaly detection and using the example dataframe in this page, I observed the same thing. For example, df in the anomaly detection tutorial page starts at ds = 2007-12-10, while anomalies_df starts at ds = 2008-01-10. However, nothing is mentioned in the documentation. It would be very helpful to add how many data points will be missing from the anomaly detection output (I guess there should be a formula for that) and what is the logic behind it.
Link
No response