Event with a window

match ← All logistic growth benchmarks

A campaign on two dates, each affecting the day before and the day after too.

Input

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

Output

Largest difference: 4.1e-11 (< 0.0001% of the input's range).

Every forecast row from both libraries.

Date	Effect Prophet	Python Prophet	Difference
2020-01-01	14.1923	14.1923	2.6e-11
2020-01-02	14.7244	14.7244	2.7e-11
2020-01-03	15.2730	15.2730	2.7e-11
2020-01-04	15.8385	15.8385	2.8e-11
2020-01-05	16.4210	16.4210	2.8e-11
2020-01-06	17.0208	17.0208	2.9e-11
2020-01-07	17.6380	17.6380	3.0e-11
2020-01-08	18.2730	18.2730	3.0e-11
2020-01-09	18.9258	18.9258	3.1e-11
2020-01-10	19.5967	19.5967	3.1e-11
2020-01-11	20.2856	20.2856	3.2e-11
2020-01-12	20.9928	20.9928	3.2e-11
2020-01-13	21.7184	21.7184	3.3e-11
2020-01-14	22.4622	22.4622	3.3e-11
2020-01-15	23.2244	23.2244	3.4e-11
2020-01-16	24.0049	24.0049	3.4e-11
2020-01-17	24.8037	24.8037	3.5e-11
2020-01-18	25.6206	25.6206	3.5e-11
2020-01-19	26.4556	26.4556	3.5e-11
2020-01-20	27.3084	27.3084	3.6e-11
2020-01-21	30.1523	30.1523	4.1e-11
2020-01-22	30.9563	30.9563	3.6e-11
2020-01-23	31.9631	31.9631	3.9e-11
2020-01-24	30.8937	30.8937	3.7e-11
2020-01-25	31.8320	31.8320	3.7e-11
2020-01-26	32.7862	32.7862	3.7e-11
2020-01-27	33.7560	33.7560	3.7e-11
2020-01-28	34.7408	34.7408	3.8e-11
2020-01-29	35.7401	35.7401	3.8e-11
2020-01-30	36.7534	36.7534	3.8e-11
2020-01-31	37.7799	37.7799	3.8e-11
2020-02-01	38.8191	38.8191	3.8e-11
2020-02-02	39.8702	39.8702	3.8e-11
2020-02-03	40.9325	40.9325	3.7e-11
2020-02-04	42.0052	42.0052	3.7e-11
2020-02-05	43.0876	43.0876	3.7e-11
2020-02-06	44.1788	44.1788	3.7e-11
2020-02-07	45.2779	45.2779	3.7e-11
2020-02-08	46.3841	46.3841	3.6e-11
2020-02-09	47.4970	47.4970	3.6e-11
2020-02-10	48.6152	48.6152	3.6e-11
2020-02-11	49.7377	49.7377	3.5e-11
2020-02-12	50.8635	50.8635	3.5e-11
2020-02-13	51.9918	51.9918	3.4e-11
2020-02-14	53.1215	53.1215	3.4e-11
2020-02-15	54.2517	54.2517	3.3e-11
2020-02-16	55.3814	55.3814	3.3e-11
2020-02-17	56.5096	56.5096	3.2e-11
2020-02-18	57.6354	57.6354	3.2e-11
2020-02-19	58.7579	58.7579	3.1e-11
2020-02-20	59.8762	59.8762	3.0e-11
2020-02-21	60.9892	60.9892	3.0e-11
2020-02-22	62.0962	62.0962	2.9e-11
2020-02-23	63.1963	63.1963	2.8e-11
2020-02-24	64.2886	64.2886	2.8e-11
2020-02-25	65.3723	65.3723	2.7e-11
2020-02-26	66.4467	66.4467	2.6e-11
2020-02-27	67.5109	67.5109	2.6e-11
2020-02-28	68.5644	68.5644	2.5e-11
2020-02-29	69.6064	69.6064	2.4e-11
2020-03-01	70.6362	70.6362	2.3e-11
2020-03-02	71.6533	71.6533	2.2e-11
2020-03-03	72.6571	72.6571	2.2e-11
2020-03-04	73.6472	73.6472	2.1e-11
2020-03-05	74.6229	74.6229	2.0e-11
2020-03-06	75.5839	75.5839	1.9e-11
2020-03-07	76.5298	76.5298	1.9e-11
2020-03-08	77.4602	77.4602	1.8e-11
2020-03-09	78.3748	78.3748	1.7e-11
2020-03-10	79.2734	79.2734	1.7e-11
2020-03-11	80.1556	80.1556	1.6e-11
2020-03-12	81.0213	81.0213	1.5e-11
2020-03-13	81.8704	81.8704	1.4e-11
2020-03-14	82.7027	82.7027	1.4e-11
2020-03-15	83.5180	83.5180	1.3e-11
2020-03-16	84.3165	84.3165	1.3e-11
2020-03-17	85.0979	85.0979	1.2e-11
2020-03-18	85.8624	85.8624	1.1e-11
2020-03-19	86.6099	86.6099	1.1e-11
2020-03-20	87.3406	87.3406	1.0e-11
2020-03-21	88.0545	88.0545	9.6e-12
2020-03-22	88.7517	88.7517	9.1e-12
2020-03-23	89.4324	89.4324	8.5e-12
2020-03-24	90.0967	90.0967	8.0e-12
2020-03-25	90.7448	90.7448	7.5e-12
2020-03-26	91.3768	91.3768	7.1e-12
2020-03-27	91.9931	91.9931	6.6e-12
2020-03-28	92.5938	92.5938	6.2e-12
2020-03-29	93.1791	93.1791	5.8e-12
2020-03-30	93.7493	93.7493	5.4e-12
2020-03-31	94.3047	94.3047	5.0e-12
2020-04-01	94.8456	94.8456	4.6e-12
2020-04-02	95.3721	95.3721	4.3e-12
2020-04-03	95.8847	95.8847	4.0e-12
2020-04-04	96.3835	96.3835	3.6e-12
2020-04-05	96.8690	96.8690	3.4e-12
2020-04-06	97.3413	97.3413	3.1e-12
2020-04-07	97.8008	97.8008	2.8e-12
2020-04-08	98.2479	98.2479	2.5e-12
2020-04-09	98.6827	98.6827	2.3e-12
2020-04-10	99.1056	99.1056	2.1e-12
2020-04-11	99.5170	99.5170	1.9e-12
2020-04-12	99.9171	99.9171	1.6e-12
2020-04-13	100.3062	100.3062	1.4e-12
2020-04-14	100.6847	100.6847	1.3e-12
2020-04-15	103.0261	103.0261	5.8e-12
2020-04-16	103.3002	103.3002	4.5e-13
2020-04-17	103.7502	103.7502	2.7e-12
2020-04-18	102.0977	102.0977	6.3e-13
2020-04-19	102.4273	102.4273	5.0e-13
2020-04-20	102.7479	102.7479	3.6e-13
2020-04-21	103.0598	103.0598	2.4e-13
2020-04-22	103.3634	103.3634	1.3e-13
2020-04-23	103.6588	103.6588	1.4e-14
2020-04-24	103.9464	103.9464	5.7e-14
2020-04-25	104.2264	104.2264	1.1e-13
2020-04-26	104.4990	104.4990	2.4e-13
2020-04-27	104.7645	104.7645	3.1e-13
2020-04-28	105.0231	105.0231	3.8e-13
2020-04-29	105.2751	105.2751	4.4e-13

Timing

Median of all measured repetitions.

Step	Effect Prophet	Python Prophet	Effect Prophet is
Fit	9.09 ms	17.9 ms	2.0× faster
Predict	925 µs	4.20 ms	4.5× faster
Save model to JSON	22 µs	656 µs	30× faster
Load model from JSON	150 µs	3.45 ms	23× faster
Load and predict in a new process	158 ms	1.49 s	9.4× 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": [
        "2020-02-08T00:00:00.000Z"
      ]
    },
    "changepointPriorScale": 0.05,
    "seasonalities": [],
    "events": [
      {
        "name": "campaign",
        "date": "2020-01-22",
        "lowerWindowDays": -1,
        "upperWindowDays": 1,
        "priorScale": 10
      },
      {
        "name": "campaign",
        "date": "2020-04-16",
        "lowerWindowDays": -1,
        "upperWindowDays": 1,
        "priorScale": 10
      }
    ],
    "regressors": []
  },
  "pythonOptimizer": {
    "algorithm": "Auto",
    "maxIterations": 10000,
    "newtonFallback": true,
    "sigFigs": 12
  }
}