Defaults, steady level

match ← All flat growth benchmarks

96 days hovering around one value. Weekly seasonality switches on automatically.

Input

benchmark/inputs/v1/flat-level.json: 96 history rows and 24 forecast rows.

Output

Largest difference: 1.2e-11 (< 0.0001% of the input's range).

Every forecast row from both libraries.

Date	Effect Prophet	Python Prophet	Difference
2020-04-06	34.9983	34.9983	1.2e-11
2020-04-07	35.0061	35.0061	1.2e-11
2020-04-08	35.0025	35.0025	1.2e-11
2020-04-09	35.0023	35.0023	1.2e-11
2020-04-10	35.0014	35.0014	1.2e-11
2020-04-11	35.0004	35.0004	1.2e-11
2020-04-12	34.9995	34.9995	1.2e-11
2020-04-13	34.9983	34.9983	1.2e-11
2020-04-14	35.0061	35.0061	1.2e-11
2020-04-15	35.0025	35.0025	1.2e-11
2020-04-16	35.0023	35.0023	1.2e-11
2020-04-17	35.0014	35.0014	1.2e-11
2020-04-18	35.0004	35.0004	1.2e-11
2020-04-19	34.9995	34.9995	1.2e-11
2020-04-20	34.9983	34.9983	1.2e-11
2020-04-21	35.0061	35.0061	1.2e-11
2020-04-22	35.0025	35.0025	1.2e-11
2020-04-23	35.0023	35.0023	1.2e-11
2020-04-24	35.0014	35.0014	1.2e-11
2020-04-25	35.0004	35.0004	1.2e-11
2020-04-26	34.9995	34.9995	1.2e-11
2020-04-27	34.9983	34.9983	1.2e-11
2020-04-28	35.0061	35.0061	1.2e-11
2020-04-29	35.0025	35.0025	1.2e-11

Timing

Median of all measured repetitions.

Step	Effect Prophet	Python Prophet	Effect Prophet is
Convert input rows	16 µs	490 µs	30× faster
Fit	2.93 ms	9.11 ms	3.1× faster
Predict	231 µs	3.02 ms	13× faster
Fit and predict	1.86 ms	11.4 ms	6.1× faster
Convert, fit and predict	1.41 ms	12.1 ms	8.6× faster
First forecast in a new process	169 ms	1.42 s	8.4× faster
Save model to JSON	24 µs	467 µs	19× faster
Load model from JSON	115 µs	2.19 ms	19× faster
Load and predict in a new process	152 ms	1.40 s	9.2× faster

Settings

The case declaration both libraries ran.

{
  "kind": "map",
  "comparison": {
    "kind": "equivalent-objective",
    "evidenceId": "flat-growth-reduced-v1"
  },
  "fitRequest": "growth-only",
  "configuration": {
    "growth": "flat",
    "scaling": "absmax",
    "seasonalityMode": "additive",
    "holidaysMode": "additive",
    "changepoints": {
      "mode": "auto",
      "count": 25,
      "range": 0.8
    },
    "changepointPriorScale": 0.05,
    "seasonalities": [
      {
        "name": "weekly",
        "periodDays": 7,
        "fourierOrder": 3,
        "priorScale": 10
      }
    ],
    "events": [],
    "regressors": []
  },
  "pythonOptimizer": {
    "algorithm": "Auto",
    "maxIterations": 10000,
    "newtonFallback": true
  }
}