Ablation.to_json()

Get the settings and results of the ablation as a JSON string.

Usage

Source

Ablation.to_json(indent=2)

The structure is the same as that of to_dict(). Saving it alongside a model’s training data makes a record of which checks mattered to the model, and of how that was measured.

Parameters

indent: int | None = 2

The number of spaces to indent nested levels by. Use None for the most compact output, on a single line.

Returns

str

The settings and results of the ablation as a JSON string.

Examples

Here’s an ablation of a table of temperature readings, with three repeats.

import json
import pointblank as pb
import polars as pl

temps = [20 + (k * 37 % 11 - 5) * 0.6 for k in range(300)]
for k in range(0, 300, 25):
    temps[k] *= 10
readings = pl.DataFrame({"temp": temps})

def evaluate(train, *, seed, variant):
    sample = train.sample(fraction=1.0, with_replacement=True, seed=seed)
    return abs(sample["temp"].mean() - 20)

validation = (
    pb.Validate(data=readings)
    .col_vals_between(columns="temp", left=-10, right=45)
    .interrogate()
)

ablation = (
    pb.Ablation(
        validation, evaluate=evaluate, metric="error", higher_is_better=False, repeats=3
    )
    .drop_each()
    .run()
)

print(json.dumps(json.loads(ablation.to_json())["settings"], indent=2))

{
  "label": null,
  "metric": "error",
  "metrics": [
    "error"
  ],
  "higher_is_better": false,
  "repeats": 3,
  "seed": 0,
  "control": "size_matched",
  "duplicates": "keep_first",
  "duplicates_order_by": null,
  "max_removed": 0.5,
  "conf_level": 0.95,
  "adjust": "holm",
  "evaluate_version": null,
  "n_rows": 300,
  "tbl_name": null
}

The settings record everything needed to repeat the ablation with the same evaluate() function, including the seed.