implement clientside signal processing

bcliang / dash-clientside-demo

Real-time Bitcoin price (BTC-USD) tracker using Dash with clientside callbacks.

MIT License

4 stars 1 forks source link

implement clientside signal processing #1

Open bcliang opened 5 years ago

bcliang commented 5 years ago

Ideas

use session store for input buffer (vs. recalculating trace each time)
simple moving average (Math.mean)
kalman filtering (kalmanjs)
generate new trace (add/extend) for filtered data on plot

bcliang commented 4 years ago

some working code exists here https://github.com/bcliang/dash-clientside-demo/tree/add-clientside-filter-example

TODO: because traces can have non-matching length (filter enabled/disabled) need a way to output CSV with aligned time. ALTERNATIVELY, just draw each trace as a separate pair of (X,Y) columns

bcliang commented 4 years ago

Made changes -- each trace has separate (x,y) paired column in CSV.

TODO: Refactor to eliminate confusing input validation performed in ui.relayout() and signal.filterSignal