Defaults, weekly pattern

match ← All flat growth benchmarks

96 days around one level with a clear weekly rhythm.

Input

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

Output

Largest difference: 1.1e-10 (< 0.0001% of the input's range).

Every forecast row from both libraries.

Date	Effect Prophet	Python Prophet	Difference
2020-04-06	33.6334	33.6334	1.0e-10
2020-04-07	33.9115	33.9115	8.2e-11
2020-04-08	35.0025	35.0025	1.0e-12
2020-04-09	36.0968	36.0968	8.4e-11
2020-04-10	36.3663	36.3663	1.1e-10
2020-04-11	35.6079	35.6079	4.8e-11
2020-04-12	34.3921	34.3921	4.7e-11
2020-04-13	33.6334	33.6334	1.0e-10
2020-04-14	33.9115	33.9115	8.2e-11
2020-04-15	35.0025	35.0025	1.0e-12
2020-04-16	36.0968	36.0968	8.4e-11
2020-04-17	36.3663	36.3663	1.1e-10
2020-04-18	35.6079	35.6079	4.8e-11
2020-04-19	34.3921	34.3921	4.7e-11
2020-04-20	33.6334	33.6334	1.0e-10
2020-04-21	33.9115	33.9115	8.2e-11
2020-04-22	35.0025	35.0025	1.0e-12
2020-04-23	36.0968	36.0968	8.4e-11
2020-04-24	36.3663	36.3663	1.1e-10
2020-04-25	35.6079	35.6079	4.8e-11
2020-04-26	34.3921	34.3921	4.7e-11
2020-04-27	33.6334	33.6334	1.0e-10
2020-04-28	33.9115	33.9115	8.2e-11
2020-04-29	35.0025	35.0025	1.0e-12

Timing

Median of all measured repetitions.

Step	Effect Prophet	Python Prophet	Effect Prophet is
Convert input rows	20 µs	467 µs	24× faster
Fit	106 ms	386 ms	3.6× faster
Predict	231 µs	3.04 ms	13× faster
Fit and predict	104 ms	395 ms	3.8× faster
Convert, fit and predict	106 ms	389 ms	3.7× faster
First forecast in a new process	292 ms	1.82 s	6.2× faster
Save model to JSON	29 µs	429 µs	15× faster
Load model from JSON	104 µs	2.15 ms	21× faster
Load and predict in a new process	152 ms	1.43 s	9.4× 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
  }
}