One hand-picked changepoint

match ← All linear growth benchmarks

32 daily rows with a changepoint on 13 January and a looser prior scale of 0.2.

Input

benchmark/inputs/v1/linear-small.json: 32 history rows and 16 forecast rows.

Output

Largest difference: 1.0e-11 (< 0.0001% of the input's range).

Every forecast row from both libraries.

Date	Effect Prophet	Python Prophet	Difference
2020-02-02	23.3806	23.3806	1.0e-11
2020-02-03	23.7230	23.7230	1.0e-11
2020-02-04	24.0653	24.0653	1.0e-11
2020-02-05	24.4076	24.4076	1.0e-11
2020-02-06	24.7500	24.7500	1.0e-11
2020-02-07	25.0923	25.0923	1.0e-11
2020-02-08	25.4346	25.4346	1.0e-11
2020-02-09	25.7769	25.7769	1.0e-11
2020-02-10	26.1193	26.1193	1.0e-11
2020-02-11	26.4616	26.4616	9.9e-12
2020-02-12	26.8039	26.8039	9.9e-12
2020-02-13	27.1463	27.1463	9.9e-12
2020-02-14	27.4886	27.4886	9.9e-12
2020-02-15	27.8309	27.8309	9.8e-12
2020-02-16	28.1733	28.1733	9.8e-12
2020-02-17	28.5156	28.5156	9.8e-12

Timing

Median of all measured repetitions.

Step	Effect Prophet	Python Prophet	Effect Prophet is
Convert input rows	6 µs	453 µs	78× faster
Fit	1.88 ms	11.1 ms	5.9× faster
Predict	191 µs	2.74 ms	14× faster
Fit and predict	1.65 ms	15.0 ms	9.1× faster
Convert, fit and predict	1.80 ms	14.8 ms	8.2× faster
First forecast in a new process	172 ms	1.48 s	8.6× faster
Save model to JSON	26 µs	350 µs	13× faster
Load model from JSON	113 µs	2.06 ms	18× faster
Load and predict in a new process	167 ms	1.44 s	8.6× faster

Settings

The case declaration both libraries ran.

{
  "kind": "linear-map",
  "comparison": {
    "kind": "equivalent-objective",
    "evidenceId": "linear-output-first-v4"
  },
  "configuration": {
    "effectMap": "explicit",
    "changepoints": {
      "mode": "explicit",
      "timestamps": [
        "2020-01-13T00:00:00.000Z"
      ]
    },
    "changepointPriorScale": 0.2,
    "seasonalities": [],
    "events": [],
    "regressors": []
  },
  "effectOptimizer": {
    "algorithm": "auto",
    "maxIterations": 10000,
    "fallback": "newton",
    "lbfgs": {
      "historySize": 5,
      "initAlpha": 0.001,
      "tolObj": 1e-12,
      "tolRelObj": 10000,
      "tolGrad": 1e-8,
      "tolRelGrad": 10000000,
      "tolParam": 1e-8
    }
  },
  "pythonOptimizer": {
    "algorithm": "Auto",
    "maxIterations": 10000,
    "newtonFallback": true,
    "sigFigs": 12
  }
}