No changepoints, loose prior, absmax scaling

match ← All logistic growth benchmarks

No changepoints but a changepoint prior scale of 10. Prophet adds a hidden placeholder changepoint here; the forecasts must still agree.

Input

benchmark/inputs/v2/logistic-basic-96.json: 96 history rows and 120 forecast rows.

Output

Largest difference: 9.0e-9 (< 0.0001% of the input's range).

Every forecast row from both libraries.

Date	Effect Prophet	Python Prophet	Difference
2020-01-01	6.3172	6.3172	2.0e-9
2020-01-02	6.5763	6.5763	2.1e-9
2020-01-03	6.8453	6.8453	2.2e-9
2020-01-04	7.1244	7.1244	2.3e-9
2020-01-05	7.4141	7.4141	2.4e-9
2020-01-06	7.7146	7.7146	2.5e-9
2020-01-07	8.0262	8.0262	2.6e-9
2020-01-08	8.3494	8.3494	2.6e-9
2020-01-09	8.6844	8.6844	2.7e-9
2020-01-10	9.0315	9.0315	2.8e-9
2020-01-11	9.3912	9.3912	2.9e-9
2020-01-12	9.7637	9.7637	3.1e-9
2020-01-13	10.1494	10.1494	3.2e-9
2020-01-14	10.5487	10.5487	3.3e-9
2020-01-15	10.9619	10.9619	3.4e-9
2020-01-16	11.3892	11.3892	3.5e-9
2020-01-17	11.8311	11.8311	3.6e-9
2020-01-18	12.2880	12.2880	3.7e-9
2020-01-19	12.7600	12.7600	3.9e-9
2020-01-20	13.2476	13.2476	4.0e-9
2020-01-21	13.7510	13.7510	4.1e-9
2020-01-22	14.2706	14.2706	4.2e-9
2020-01-23	14.8067	14.8067	4.4e-9
2020-01-24	15.3595	15.3595	4.5e-9
2020-01-25	15.9293	15.9293	4.6e-9
2020-01-26	16.5163	16.5163	4.8e-9
2020-01-27	17.1209	17.1209	4.9e-9
2020-01-28	17.7432	17.7432	5.1e-9
2020-01-29	18.3834	18.3834	5.2e-9
2020-01-30	19.0418	19.0418	5.3e-9
2020-01-31	19.7184	19.7184	5.5e-9
2020-02-01	20.4134	20.4134	5.6e-9
2020-02-02	21.1269	21.1269	5.8e-9
2020-02-03	21.8591	21.8591	5.9e-9
2020-02-04	22.6098	22.6098	6.1e-9
2020-02-05	23.3792	23.3792	6.2e-9
2020-02-06	24.1672	24.1672	6.4e-9
2020-02-07	24.9739	24.9739	6.5e-9
2020-02-08	25.7990	25.7990	6.7e-9
2020-02-09	26.6425	26.6425	6.8e-9
2020-02-10	27.5042	27.5042	7.0e-9
2020-02-11	28.3840	28.3840	7.1e-9
2020-02-12	29.2815	29.2815	7.2e-9
2020-02-13	30.1965	30.1965	7.4e-9
2020-02-14	31.1287	31.1287	7.5e-9
2020-02-15	32.0776	32.0776	7.6e-9
2020-02-16	33.0430	33.0430	7.8e-9
2020-02-17	34.0243	34.0243	7.9e-9
2020-02-18	35.0210	35.0210	8.0e-9
2020-02-19	36.0326	36.0326	8.1e-9
2020-02-20	37.0585	37.0585	8.2e-9
2020-02-21	38.0982	38.0982	8.3e-9
2020-02-22	39.1508	39.1508	8.4e-9
2020-02-23	40.2158	40.2158	8.5e-9
2020-02-24	41.2924	41.2924	8.6e-9
2020-02-25	42.3798	42.3798	8.7e-9
2020-02-26	43.4772	43.4772	8.7e-9
2020-02-27	44.5838	44.5838	8.8e-9
2020-02-28	45.6988	45.6988	8.9e-9
2020-02-29	46.8211	46.8211	8.9e-9
2020-03-01	47.9500	47.9500	8.9e-9
2020-03-02	49.0845	49.0845	9.0e-9
2020-03-03	50.2236	50.2236	9.0e-9
2020-03-04	51.3663	51.3663	9.0e-9
2020-03-05	52.5118	52.5118	9.0e-9
2020-03-06	53.6591	53.6591	9.0e-9
2020-03-07	54.8070	54.8070	9.0e-9
2020-03-08	55.9547	55.9547	9.0e-9
2020-03-09	57.1012	57.1012	9.0e-9
2020-03-10	58.2456	58.2456	9.0e-9
2020-03-11	59.3868	59.3868	8.9e-9
2020-03-12	60.5238	60.5238	8.9e-9
2020-03-13	61.6559	61.6559	8.8e-9
2020-03-14	62.7821	62.7821	8.8e-9
2020-03-15	63.9014	63.9014	8.7e-9
2020-03-16	65.0131	65.0131	8.6e-9
2020-03-17	66.1163	66.1163	8.5e-9
2020-03-18	67.2101	67.2101	8.5e-9
2020-03-19	68.2939	68.2939	8.4e-9
2020-03-20	69.3669	69.3669	8.3e-9
2020-03-21	70.4284	70.4284	8.2e-9
2020-03-22	71.4777	71.4777	8.1e-9
2020-03-23	72.5143	72.5143	7.9e-9
2020-03-24	73.5375	73.5375	7.8e-9
2020-03-25	74.5468	74.5468	7.7e-9
2020-03-26	75.5417	75.5417	7.6e-9
2020-03-27	76.5217	76.5217	7.5e-9
2020-03-28	77.4865	77.4865	7.3e-9
2020-03-29	78.4356	78.4356	7.2e-9
2020-03-30	79.3687	79.3687	7.0e-9
2020-03-31	80.2856	80.2856	6.9e-9
2020-04-01	81.1860	81.1860	6.8e-9
2020-04-02	82.0696	82.0696	6.6e-9
2020-04-03	82.9363	82.9363	6.5e-9
2020-04-04	83.7859	83.7859	6.3e-9
2020-04-05	84.6184	84.6184	6.2e-9
2020-04-06	85.4337	85.4337	6.0e-9
2020-04-07	86.2318	86.2318	5.9e-9
2020-04-08	87.0125	87.0125	5.8e-9
2020-04-09	87.7761	87.7761	5.6e-9
2020-04-10	88.5224	88.5224	5.5e-9
2020-04-11	89.2517	89.2517	5.3e-9
2020-04-12	89.9639	89.9639	5.2e-9
2020-04-13	90.6593	90.6593	5.0e-9
2020-04-14	91.3380	91.3380	4.9e-9
2020-04-15	92.0002	92.0002	4.8e-9
2020-04-16	92.6460	92.6460	4.6e-9
2020-04-17	93.2757	93.2757	4.5e-9
2020-04-18	93.8895	93.8895	4.4e-9
2020-04-19	94.4876	94.4876	4.2e-9
2020-04-20	95.0703	95.0703	4.1e-9
2020-04-21	95.6378	95.6378	4.0e-9
2020-04-22	96.1904	96.1904	3.9e-9
2020-04-23	96.7285	96.7285	3.7e-9
2020-04-24	97.2522	97.2522	3.6e-9
2020-04-25	97.7619	97.7619	3.5e-9
2020-04-26	98.2579	98.2579	3.4e-9
2020-04-27	98.7404	98.7404	3.3e-9
2020-04-28	99.2099	99.2099	3.2e-9
2020-04-29	99.6666	99.6666	3.1e-9

Timing

Median of all measured repetitions.

Step	Effect Prophet	Python Prophet	Effect Prophet is
Fit	116 ms	128 ms	1.1× faster
Predict	719 µs	3.12 ms	4.3× faster
Save model to JSON	30 µs	427 µs	14× faster
Load model from JSON	150 µs	2.09 ms	14× faster
Load and predict in a new process	154 ms	1.54 s	10× faster

Settings

The case declaration both libraries ran.

{
  "kind": "map",
  "comparison": {
    "kind": "equivalent-objective",
    "evidenceId": "logistic-python-reconciliation-v1"
  },
  "configuration": {
    "growth": "logistic",
    "scaling": "absmax",
    "seasonalityMode": "additive",
    "holidaysMode": "additive",
    "changepoints": {
      "mode": "explicit",
      "timestamps": []
    },
    "changepointPriorScale": 10,
    "seasonalities": [],
    "events": [],
    "regressors": []
  },
  "pythonOptimizer": {
    "algorithm": "Auto",
    "maxIterations": 10000,
    "newtonFallback": true,
    "sigFigs": 12
  }
}