Workflow

WRITE ME

Batch

digraph idq_batch { labeljust = "r"; label="idq-batch" rankdir=TB; graph [fontname="helvetica", fontsize=24]; edge [ fontname="helvetica", fontsize=10 ]; node [fontname="helvetica", shape=box, fontsize=11]; style=rounded; labeljust = "r"; fontsize = 14; DataSrc [label="auxiliary features"]; Train [label="batch.train"]; Evaluate [label="batch.evaluate"]; Calibrate [label="batch.calibrate"]; Timeseries [label="batch.timeseries"]; DataSrc -> Train; Train -> Evaluate [label="model"]; DataSrc -> Evaluate; Evaluate -> Calibrate [label="quiver"]; DataSrc -> Timeseries; Train -> Timeseries [label="model"]; Calibrate -> Timeseries [label="calibration map"]; PGlitch [label="p(glitch)"] Timeseries -> PGlitch; }

TODO:

  • Describe flow chart in more detail

  • Describe how data will be processed

  • Describe how distributed processes will communicate with one another.

  • Add round-robin based training workflow

Stream

digraph idq_streaming { labeljust = "r"; label="idq-stream" rankdir=TB; graph [fontname="helvetica", fontsize=24]; edge [ fontname="helvetica", fontsize=10 ]; node [fontname="helvetica", shape=box, fontsize=11]; style=rounded; labeljust = "r"; fontsize = 14; DataSrc [label="auxiliary features"]; Train [label="stream.train"]; Evaluate [label="stream.evaluate"]; Calibrate [label="stream.calibrate"]; Timeseries [label="stream.timeseries"]; DataSrc -> Train; Train -> Evaluate [label="model"]; DataSrc -> Evaluate; Evaluate -> Calibrate [label="quiver"]; DataSrc -> Timeseries; Train -> Timeseries [label="model"]; Calibrate -> Timeseries [label="calibration map"]; PGlitch [label="p(glitch)"] Timeseries -> PGlitch; }

Batch vs Stream modes

WRITE ME

Tasks

WRITE ME

Training

WRITE ME

Evaluation

WRITE ME

Calibration

WRITE ME

Timeseries

WRITE ME