Repository navigation
Methods Filter Bank
FilterBank (many streaming filters driven in lockstep) and
FusedChiSquareDetector (a pooled multi-stream fault test built on them)
live in dtfit_experimental.streaming -- experiment harness tooling, not
part of the stable dtfit surface. See
the experimental adaptations API.
The stable equivalent of the fused test is summing several
ImageFilter instances' nis_: each is
chi-square under its model, so the sum is chi-square with the summed
degrees of freedom, giving the pooled test more degrees of freedom and
power than any one filter's innovation alone when the change is spread
evenly over the streams; on the three-axis damping fault of experiment
21, carried mostly by one axis, per-axis tests on every sample were as
fast. See
several streams.
On a flag the detector re-arms every filter: inflate= multiplies each
P by the given factor and keeps the windows, rearm=True calls each
filter's rearm() instead, which applies the filter's own drift_reset
and collapses its adaptive window. A bank of one filter is the per-sample
innovation test on a single stream.
Published automatically from the repo's wiki/ directory by the wiki workflow. All pages are maintained directly in wiki/.
Start here
Guides
Examples
- Overview
- 01 Quickstart
- 02 Fitting methods
- 03 Models & auto
- 04 sklearn estimator
- 05 Streaming
- 06 Scaling
- 07 Diagnostics
- 08 Stochastic
API reference
Methods (math)
Experimental
Validation