India with explicit known-gap opt-in

match ← All flat growth benchmarks

A 2036 history and 2037 forecasts use the pinned partial provider result with allow-known-gaps. The omitted table-backed holidays are not claimed complete.

Input

benchmark/inputs/v5/country-partial-96.json: 96 history rows and 3 forecast rows.

Output

Largest difference: 3.2e-12 (< 0.0001% of the input's range).

Every forecast row from both libraries.

Date	Effect Prophet	Python Prophet	Difference
2037-01-01	20.0510	20.0510	3.0e-12
2037-01-26	19.9291	19.9291	2.9e-12
2037-08-15	20.0510	20.0510	3.2e-12

Timing

Median of all measured repetitions.

Step	Effect Prophet	Python Prophet	Effect Prophet is
Convert input rows	15 µs	446 µs	30× faster
Fit	3.64 ms	15.2 ms	4.2× faster
Predict	208 µs	4.85 ms	23× faster
Fit and predict	3.40 ms	20.3 ms	6.0× faster
Convert, fit and predict	3.10 ms	20.9 ms	6.7× faster
First forecast in a new process	179 ms	1.59 s	8.9× faster
Save model to JSON	120 µs	510 µs	4.2× faster
Load model from JSON	1.27 ms	3.06 ms	2.4× faster
Load and predict in a new process	163 ms	1.58 s	9.7× faster

Settings

The case declaration both libraries ran.

{
  "kind": "map",
  "comparison": {
    "kind": "equivalent-objective",
    "evidenceId": "country-calendar-public-v1"
  },
  "configuration": {
    "growth": "flat",
    "countryHolidays": {
      "country": "IN",
      "coverage": "allow-known-gaps",
      "priorScale": 10
    },
    "scaling": "absmax",
    "seasonalityMode": "additive",
    "holidaysMode": "additive",
    "changepoints": {
      "mode": "auto",
      "count": 25,
      "range": 0.8
    },
    "changepointPriorScale": 0.05,
    "seasonalities": [],
    "events": [],
    "regressors": []
  },
  "pythonOptimizer": {
    "algorithm": "Auto",
    "maxIterations": 10000,
    "newtonFallback": true,
    "sigFigs": 12
  }
}