SQLiteTracker

class sklearn_evaluation.SQLiteTracker(path: str)

A simple experiment tracker using SQLite

Click here to see the user guide.

Parameters

path – Database location

comment(uuid, comment)

Add a comment to an experiment given its uuid

insert(uuid, parameters)

Insert a new experiment

new()

Create a new experiment, returns a uuid

query(code)

Query the database, returns a pandas.DataFrame

Examples

>>> from sklearn_evaluation import SQLiteTracker
>>> tracker = SQLiteTracker(':memory:') # example in-memory db
>>> tracker.insert('my_uuid', {'a': 1})
>>> df = tracker.query(
... "SELECT uuid, json_extract(parameters, '$.a') FROM experiments")
recent(n=5, normalize=False)

Get most recent experiments as a pandas.DataFrame

update(uuid, parameters)

Update the parameters of an empty experiment given its uuid