Switchable weekly pattern

match ← All logistic growth benchmarks

A weekly pattern that only applies on rows marked active.

Input

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

Output

Largest difference: 4.8e-11 (< 0.0001% of the input's range).

Every forecast row from both libraries.

Date	Effect Prophet	Python Prophet	Difference
2020-01-01	14.1882	14.1882	5.7e-12
2020-01-02	15.3458	15.3458	6.3e-12
2020-01-03	16.0443	16.0443	6.4e-12
2020-01-04	16.1815	16.1815	6.9e-12
2020-01-05	16.0672	16.0672	7.6e-12
2020-01-06	17.0165	17.0165	8.5e-12
2020-01-07	17.0171	17.0171	9.6e-12
2020-01-08	18.2694	18.2694	1.0e-11
2020-01-09	19.5471	19.5471	1.1e-11
2020-01-10	20.3679	20.3679	1.1e-11
2020-01-11	20.2813	20.2813	1.2e-11
2020-01-12	20.6390	20.6390	1.2e-11
2020-01-13	20.9313	20.9313	1.4e-11
2020-01-14	21.8413	21.8413	1.5e-11
2020-01-15	23.2208	23.2208	1.6e-11
2020-01-16	24.0007	24.0007	1.6e-11
2020-01-17	25.5750	25.5750	1.6e-11
2020-01-18	25.9637	25.9637	1.7e-11
2020-01-19	26.1020	26.1020	1.8e-11
2020-01-20	26.5217	26.5217	2.0e-11
2020-01-21	28.1750	28.1750	2.0e-11
2020-01-22	29.0637	29.0637	2.2e-11
2020-01-23	30.5937	30.5937	2.2e-11
2020-01-24	31.6656	31.6656	2.3e-11
2020-01-25	32.1756	32.1756	2.4e-11
2020-01-26	32.7828	32.7828	2.5e-11
2020-01-27	32.9699	32.9699	2.6e-11
2020-01-28	34.1210	34.1210	2.8e-11
2020-01-29	35.7377	35.7377	2.9e-11
2020-01-30	37.3759	37.3759	2.9e-11
2020-01-31	37.7771	37.7771	3.0e-11
2020-02-01	39.1636	39.1636	3.1e-11
2020-02-02	39.5181	39.5181	3.2e-11
2020-02-03	40.1474	40.1474	3.3e-11
2020-02-04	41.3864	41.3864	3.5e-11
2020-02-05	43.0855	43.0855	3.5e-11
2020-02-06	44.8024	44.8024	3.6e-11
2020-02-07	46.0517	46.0517	3.7e-11
2020-02-08	46.7298	46.7298	3.7e-11
2020-02-09	47.1460	47.1460	3.9e-11
2020-02-10	48.6138	48.6138	4.0e-11
2020-02-11	49.1197	49.1197	4.1e-11
2020-02-12	50.8629	50.8629	4.2e-11
2020-02-13	52.6161	52.6161	4.2e-11
2020-02-14	53.8958	53.8958	4.2e-11
2020-02-15	54.2505	54.2505	4.3e-11
2020-02-16	55.0307	55.0307	4.4e-11
2020-02-17	55.7258	55.7258	4.5e-11
2020-02-18	57.0178	57.0178	4.5e-11
2020-02-19	58.7577	58.7577	4.6e-11
2020-02-20	59.8753	59.8753	4.6e-11
2020-02-21	61.7639	61.7639	4.6e-11
2020-02-22	62.4427	62.4427	4.6e-11
2020-02-23	62.8461	62.8461	4.7e-11
2020-02-24	63.5052	63.5052	4.7e-11
2020-02-25	65.3718	65.3718	4.7e-11
2020-02-26	66.4469	66.4469	4.8e-11
2020-02-27	68.1360	68.1360	4.8e-11
2020-02-28	69.3395	69.3395	4.8e-11
2020-02-29	69.9533	69.9533	4.7e-11
2020-03-01	70.6359	70.6359	4.8e-11
2020-03-02	70.8704	70.8704	4.8e-11
2020-03-03	72.0404	72.0404	4.8e-11
2020-03-04	73.6478	73.6478	4.8e-11
2020-03-05	75.2484	75.2484	4.7e-11
2020-03-06	75.5839	75.5839	4.7e-11
2020-03-07	76.8770	76.8770	4.6e-11
2020-03-08	77.1107	77.1107	4.6e-11
2020-03-09	77.5922	77.5922	4.6e-11
2020-03-10	78.6569	78.6569	4.6e-11
2020-03-11	80.1558	80.1558	4.5e-11
2020-03-12	81.6471	81.6471	4.5e-11
2020-03-13	82.6462	82.6462	4.4e-11
2020-03-14	83.0502	83.0502	4.3e-11
2020-03-15	83.1688	83.1688	4.3e-11
2020-03-16	84.3168	84.3168	4.3e-11
2020-03-17	84.4817	84.4817	4.3e-11
2020-03-18	85.8635	85.8635	4.2e-11
2020-03-19	87.2359	87.2359	4.1e-11
2020-03-20	88.1166	88.1166	4.0e-11
2020-03-21	88.0550	88.0550	4.0e-11
2020-03-22	88.4027	88.4027	3.9e-11
2020-03-23	88.6502	88.6502	3.8e-11
2020-03-24	89.4806	89.4806	3.8e-11
2020-03-25	90.7460	90.7460	3.7e-11
2020-03-26	91.3774	91.3774	3.6e-11
2020-03-27	92.7692	92.7692	3.5e-11
2020-03-28	92.9416	92.9416	3.4e-11
2020-03-29	92.8301	92.8301	3.4e-11
2020-03-30	92.9672	92.9672	3.4e-11
2020-03-31	94.3053	94.3053	3.3e-11
2020-04-01	94.8469	94.8469	3.2e-11
2020-04-02	95.9983	95.9983	3.1e-11
2020-04-03	96.6608	96.6608	3.0e-11
2020-04-04	96.7314	96.7314	2.9e-11
2020-04-05	96.8696	96.8696	2.9e-11
2020-04-06	96.5592	96.5592	2.9e-11
2020-04-07	97.1848	97.1848	2.8e-11
2020-04-08	98.2492	98.2492	2.7e-11
2020-04-09	99.3088	99.3088	2.6e-11
2020-04-10	99.1063	99.1063	2.6e-11
2020-04-11	99.8648	99.8648	2.5e-11
2020-04-12	99.5682	99.5682	2.4e-11
2020-04-13	99.5241	99.5241	2.4e-11
2020-04-14	100.0687	100.0687	2.4e-11
2020-04-15	101.0534	101.0534	2.2e-11
2020-04-16	102.0369	102.0369	2.2e-11
2020-04-17	102.5351	102.5351	2.1e-11
2020-04-18	102.4455	102.4455	2.0e-11
2020-04-19	102.0783	102.0783	2.0e-11
2020-04-20	102.7484	102.7484	2.0e-11
2020-04-21	102.4437	102.4437	1.9e-11
2020-04-22	103.3646	103.3646	1.9e-11
2020-04-23	104.2849	104.2849	1.8e-11
2020-04-24	104.7224	104.7224	1.7e-11
2020-04-25	104.2269	104.2269	1.7e-11
2020-04-26	104.1499	104.1499	1.6e-11
2020-04-27	103.9822	103.9822	1.6e-11
2020-04-28	104.4070	104.4070	1.6e-11
2020-04-29	105.2763	105.2763	1.5e-11

Timing

Median of all measured repetitions.

Step	Effect Prophet	Python Prophet	Effect Prophet is
Fit	16.2 ms	28.3 ms	1.7× faster
Predict	1.08 ms	3.55 ms	3.3× faster
Save model to JSON	28 µs	481 µs	17× faster
Load model from JSON	148 µs	2.50 ms	17× faster
Load and predict in a new process	169 ms	1.49 s	8.8× 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": [
      {
        "name": "weekly-custom",
        "periodDays": 7,
        "fourierOrder": 3,
        "priorScale": 10,
        "conditionName": "active"
      }
    ],
    "events": [],
    "regressors": []
  },
  "pythonOptimizer": {
    "algorithm": "Auto",
    "maxIterations": 10000,
    "newtonFallback": true,
    "sigFigs": 12
  }
}