Values far from zero, minmax scaling

match ← All linear growth benchmarks

The same data scaled to the 0–1 range using its minimum and maximum.

Input

benchmark/inputs/v1/linear-offset-scaling.json: 96 history rows and 24 forecast rows.

Output

Largest difference: 8.5e-13 (< 0.0001% of the input's range).

Every forecast row from both libraries.

Date	Effect Prophet	Python Prophet	Difference
2020-04-06	110.0815	110.0815	7.2e-13
2020-04-07	110.2542	110.2542	4.4e-13
2020-04-08	110.5692	110.5692	2.8e-13
2020-04-09	110.8822	110.8822	3.7e-13
2020-04-10	111.0509	111.0509	6.0e-13
2020-04-11	111.0351	111.0351	8.2e-13
2020-04-12	110.9358	110.9358	8.5e-13
2020-04-13	110.9227	110.9227	6.8e-13
2020-04-14	111.0954	111.0954	4.1e-13
2020-04-15	111.4104	111.4104	2.6e-13
2020-04-16	111.7234	111.7234	3.4e-13
2020-04-17	111.8921	111.8921	5.4e-13
2020-04-18	111.8763	111.8763	7.7e-13
2020-04-19	111.7770	111.7770	8.1e-13
2020-04-20	111.7639	111.7639	6.3e-13
2020-04-21	111.9366	111.9366	3.6e-13
2020-04-22	112.2516	112.2516	2.0e-13
2020-04-23	112.5646	112.5646	2.7e-13
2020-04-24	112.7333	112.7333	5.1e-13
2020-04-25	112.7175	112.7175	7.2e-13
2020-04-26	112.6182	112.6182	7.5e-13
2020-04-27	112.6051	112.6051	5.8e-13
2020-04-28	112.7778	112.7778	3.1e-13
2020-04-29	113.0928	113.0928	1.8e-13

Timing

Median of all measured repetitions.

Step	Effect Prophet	Python Prophet	Effect Prophet is
Convert input rows	16 µs	437 µs	28× faster
Fit	5.41 ms	24.3 ms	4.5× faster
Predict	180 µs	3.14 ms	17× faster
Fit and predict	5.28 ms	26.7 ms	5.1× faster
Convert, fit and predict	5.65 ms	26.9 ms	4.8× faster
First forecast in a new process	179 ms	1.49 s	8.3× faster
Save model to JSON	20 µs	409 µs	21× faster
Load model from JSON	110 µs	2.15 ms	20× faster
Load and predict in a new process	153 ms	1.43 s	9.3× faster

Settings

The case declaration both libraries ran.

{
  "kind": "map",
  "comparison": {
    "kind": "equivalent-objective",
    "evidenceId": "linear-output-first-v4"
  },
  "configuration": {
    "growth": "linear",
    "scaling": "minmax",
    "seasonalityMode": "additive",
    "holidaysMode": "additive",
    "changepoints": {
      "mode": "explicit",
      "timestamps": [
        "2020-02-10T00:00:00.000Z"
      ]
    },
    "changepointPriorScale": 0.2,
    "seasonalities": [
      {
        "name": "weekly-custom",
        "periodDays": 7,
        "fourierOrder": 2,
        "priorScale": 10
      }
    ],
    "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
  }
}