Repeated timestamps, hand-picked changepoint

match ← All linear growth benchmarks

Some dates appear several times with different values. Every row counts as its own measurement.

Input

benchmark/inputs/v1/training-duplicates.json: 96 history rows and 8 forecast rows.

Output

Largest difference: 1.7e-11 (< 0.0001% of the input's range).

Every forecast row from both libraries.

Date	Effect Prophet	Python Prophet	Difference
2020-01-25	16.2514	16.2514	1.4e-11
2020-01-26	16.4832	16.4832	1.4e-11
2020-01-27	16.7149	16.7149	1.5e-11
2020-01-28	16.9467	16.9467	1.5e-11
2020-01-29	17.1784	17.1784	1.6e-11
2020-01-30	17.4102	17.4102	1.6e-11
2020-01-31	17.6420	17.6420	1.6e-11
2020-02-01	17.8737	17.8737	1.7e-11

Timing

Median of all measured repetitions.

Step	Effect Prophet	Python Prophet	Effect Prophet is
Fit	3.96 ms	19.8 ms	5.0× faster
Predict	178 µs	2.61 ms	15× faster
Save model to JSON	18 µs	405 µs	22× faster
Load model from JSON	137 µs	2.05 ms	15× faster
Load and predict in a new process	155 ms	1.55 s	10× faster

Settings

The case declaration both libraries ran.

{
  "kind": "map",
  "comparison": {
    "kind": "equivalent-objective",
    "evidenceId": "linear-output-first-v4"
  },
  "configuration": {
    "growth": "linear",
    "scaling": "absmax",
    "seasonalityMode": "additive",
    "holidaysMode": "additive",
    "changepoints": {
      "mode": "explicit",
      "timestamps": [
        "2020-01-17T00:00:00.000Z"
      ]
    },
    "changepointPriorScale": 0.05,
    "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
  }
}