Discussions
Is there another model besides timegpt-1 for anomaly detection?
13 days ago by Andreas
Hey!
I've been looking into the docs for models used for anomaly detection apart from timegpt-1, however I couldn't find something relevant. So, I tried the following:
anomalies_df = nixtla_client.detect_anomalies(
df = pharmacy_drug_complete_ts[column_subset],
freq = 'D',
date_features_to_one_hot = False,
time_col = 'order_date',
target_col = 'amount_requested',
model='timegpt-1'
)
anomalies_df = nixtla_client.detect_anomalies(
df = pharmacy_drug_complete_ts[column_subset],
freq = 'D',
date_features_to_one_hot = False,
time_col = 'order_date',
target_col = 'amount_requested',
model='timegpt-1-long-horizon'
)
and it seems to be working. Can someone verify? is there some documentation somewhere?
Thank you!