Anomaly Detection with Dogelog Payer (Was: ANN: Dogelog Player 1.2.4)

Dogelog Player is a Prolog system written
in 100% Prolog. We discovered that XSB Prolog
provides an interesting predicate unnumbervars/3
and did a rework of library(aggregate). We show
how the library can be used for anomaly
detection in time series.
Together with the non-ISO core extension call_nth/2
from library(sequence) one can quickly define
sliding windows computations in Prolog itself.
We compute moving variance and moving delta for
actual bitcoin market prices. Using both indicators
we can identify an abnormally in november.
Anomaly Detection with Dogelog Payer
https://twitter.com/dogelogch/status/1860129082723467710
Anomaly Detection with Dogelog Payer
https://www.facebook.com/groups/dogelog
Mild Shock schrieb:

Dear All,
 We are happy to announce a new edition of
the Dogelog player:
 - New library(misc/dict):
   We have moved some predicates from the library(misc/json)
into our own library(misc/dict).  We made sure that the
data structure of the Prolog dicts now follows the
input order semantics.
 - New library(util/hash):
   To enable the implementation of hash tables, Novacore
now has a term_hash/2 predicate.  The implementation
differs from the Prolog dict in that change_arg/3 is
used for destructive updates.  We have also dispensed
with the input order.
 - Improved library(sequence):
   Hash tables can easily be used to model sets of keys,
for example by using the value "true" as an indicator.
This allowed us to re-implement the distinct/1 predicate.
The use of numbervars/3 ensures that key variants
are identified.
 Have Fun!
 Jan Burse, http://www.xlog.ch/ , 02.11.2024