presc.report package

presc.report.runner module

class presc.report.runner.Context(store_dir='.')[source]

Bases: object

Persistent data store for sharing report inputs across notebooks.

Note that the store implementation does not support concurrent access. It is up to the caller to ensure that multiple instances each have a unique store location.

store_dir

The dir to contain the data store, implemented as one or more database files. If not specified, defaults to the current working dir.

Type: str

property config

property model

store_inputs(model=None, test_dataset=None, train_dataset=None, config=None)[source]

Write the report inputs to the data store.

Any existing values will be overwritten.

Parameters

model (presc.model.ClassificationModel) – A ClassificationModel instance
test_dataset (presc.dataset.Dataset) – A Dataset instance
train_dataset (presc.dataset.Dataset) – A Dataset instance
config (dict) – A dict of config options

property test_dataset

property train_dataset

class presc.report.runner.ReportRunner(output_path='.', execution_path=None, config_filepath=None)[source]

Bases: object

Main entrypoint to run the PRESC report for the given modeling inputs.

The report is written to <output_path>/presc_report. If this dir already exists, it will be overwritten.

To generate the report: ` pr = ReportRunner() pr.run(...) `

The path to the report is accessible as pr.report_html and will open in the default browser by calling pr.open()

output_path

Path to the dir where the report will be written. If not specified, defaults to the current working dir.

Type: str

execution_path

Path from which the report is built. If not specified, a temporary dir is used.

Type: str

config_filepath

Path to a YAML file listing overrides to the default config options.

Type: str

clean()[source]: Remove artifacts from a previous run, if any.

open()[source]: Open the report in the default web browser.

property report_html: The main page of the HTML report.

run(model, test_dataset, train_dataset=None, settings=None, clean=True)[source]

Runs the PRESC report for the given modeling inputs.

The report is written to <output_path>/presc_report. If this dir already exists, it will be overwritten.

Parameters

model (presc.model.ClassificationModel) – A pre-trained ClassificationModel instance to evaluate
test_dataset (presc.dataset.Dataset) – A test Dataset instance used to evaluate model performance
train_dataset (presc.dataset.Dataset) – The Dataset instance used to train the model. This is not required for every evaluation.
settings (dict) – A dict specifying option values to override report settings, eg. {“report.title”: “My Report”}.
clean (book) – Should previous outputs be cleaned? Default: True