Multiplicative regressor

match ← All logistic growth benchmarks

A promotion regressor that scales the trend.

Input

benchmark/inputs/v2/logistic-regressor-multiplicative-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	14.1877	14.1877	3.2e-11
2020-01-02	14.7198	14.7198	3.2e-11
2020-01-03	15.7273	15.7273	3.5e-11
2020-01-04	15.8338	15.8338	3.4e-11
2020-01-05	16.4163	16.4163	3.5e-11
2020-01-06	17.0161	17.0161	3.5e-11
2020-01-07	17.6333	17.6333	3.6e-11
2020-01-08	18.2683	18.2683	3.7e-11
2020-01-09	19.4898	19.4898	3.9e-11
2020-01-10	19.5919	19.5919	3.8e-11
2020-01-11	20.2809	20.2809	3.8e-11
2020-01-12	20.9881	20.9881	3.9e-11
2020-01-13	21.7136	21.7136	4.0e-11
2020-01-14	22.4575	22.4575	4.0e-11
2020-01-15	23.9176	23.9176	4.3e-11
2020-01-16	24.0003	24.0003	4.1e-11
2020-01-17	24.7991	24.7991	4.2e-11
2020-01-18	25.6161	25.6161	4.2e-11
2020-01-19	26.4512	26.4512	4.3e-11
2020-01-20	27.3041	27.3041	4.3e-11
2020-01-21	29.0215	29.0215	4.6e-11
2020-01-22	29.0627	29.0627	4.4e-11
2020-01-23	29.9678	29.9678	4.4e-11
2020-01-24	30.8897	30.8897	4.5e-11
2020-01-25	31.8281	31.8281	4.5e-11
2020-01-26	32.7824	32.7824	4.5e-11
2020-01-27	34.7668	34.7668	4.8e-11
2020-01-28	34.7373	34.7373	4.5e-11
2020-01-29	35.7368	35.7368	4.6e-11
2020-01-30	36.7501	36.7501	4.6e-11
2020-01-31	37.7768	37.7768	4.6e-11
2020-02-01	38.8161	38.8161	4.6e-11
2020-02-02	41.0656	41.0656	4.8e-11
2020-02-03	40.9299	40.9299	4.5e-11
2020-02-04	42.0028	42.0028	4.5e-11
2020-02-05	43.0854	43.0854	4.5e-11
2020-02-06	44.1768	44.1768	4.5e-11
2020-02-07	45.2761	45.2761	4.4e-11
2020-02-08	47.7765	47.7765	4.7e-11
2020-02-09	47.4954	47.4954	4.4e-11
2020-02-10	48.6136	48.6136	4.3e-11
2020-02-11	49.7361	49.7361	4.3e-11
2020-02-12	50.8620	50.8620	4.2e-11
2020-02-13	51.9902	51.9902	4.2e-11
2020-02-14	54.7165	54.7165	4.4e-11
2020-02-15	54.2501	54.2501	4.0e-11
2020-02-16	55.3799	55.3799	4.0e-11
2020-02-17	56.5081	56.5081	3.9e-11
2020-02-18	57.6340	57.6340	3.8e-11
2020-02-19	58.7565	58.7565	3.8e-11
2020-02-20	61.6743	61.6743	4.0e-11
2020-02-21	60.9879	60.9879	3.6e-11
2020-02-22	62.0949	62.0949	3.5e-11
2020-02-23	63.1950	63.1950	3.4e-11
2020-02-24	64.2873	64.2873	3.3e-11
2020-02-25	65.3711	65.3711	3.3e-11
2020-02-26	68.4425	68.4425	3.5e-11
2020-02-27	67.5098	67.5098	3.1e-11
2020-02-28	68.5632	68.5632	3.0e-11
2020-02-29	69.6052	69.6052	2.9e-11
2020-03-01	70.6351	70.6351	2.8e-11
2020-03-02	71.6522	71.6522	2.7e-11
2020-03-03	74.8398	74.8398	3.0e-11
2020-03-04	73.6462	73.6462	2.5e-11
2020-03-05	74.6219	74.6219	2.4e-11
2020-03-06	75.5830	75.5830	2.3e-11
2020-03-07	76.5289	76.5289	2.2e-11
2020-03-08	77.4593	77.4593	2.2e-11
2020-03-09	80.7295	80.7295	2.4e-11
2020-03-10	79.2726	79.2726	2.0e-11
2020-03-11	80.1548	80.1548	1.9e-11
2020-03-12	81.0206	81.0206	1.8e-11
2020-03-13	81.8697	81.8697	1.7e-11
2020-03-14	82.7020	82.7020	1.7e-11
2020-03-15	86.0275	86.0275	2.0e-11
2020-03-16	84.3158	84.3158	1.5e-11
2020-03-17	85.0973	85.0973	1.4e-11
2020-03-18	85.8618	85.8618	1.3e-11
2020-03-19	86.6094	86.6094	1.3e-11
2020-03-20	87.3401	87.3401	1.2e-11
2020-03-21	90.7005	90.7005	1.5e-11
2020-03-22	88.7512	88.7512	1.1e-11
2020-03-23	89.4319	89.4319	1.0e-11
2020-03-24	90.0963	90.0963	9.5e-12
2020-03-25	90.7444	90.7444	8.9e-12
2020-03-26	91.3764	91.3764	8.4e-12
2020-03-27	94.7576	94.7576	1.2e-11
2020-03-28	92.5934	92.5934	7.3e-12
2020-03-29	93.1788	93.1788	6.8e-12
2020-03-30	93.7490	93.7490	6.3e-12
2020-03-31	94.3044	94.3044	5.9e-12
2020-04-01	94.8453	94.8453	5.4e-12
2020-04-02	98.2383	98.2383	8.9e-12
2020-04-03	95.8844	95.8844	4.6e-12
2020-04-04	96.3833	96.3833	4.2e-12
2020-04-05	96.8687	96.8687	3.9e-12
2020-04-06	97.3411	97.3411	3.5e-12
2020-04-07	97.8006	97.8006	3.2e-12
2020-04-08	101.2005	101.2005	6.9e-12
2020-04-09	98.6825	98.6825	2.6e-12
2020-04-10	99.1055	99.1055	2.3e-12
2020-04-11	99.5169	99.5169	2.1e-12
2020-04-12	99.9170	99.9170	1.8e-12
2020-04-13	100.3061	100.3061	1.6e-12
2020-04-14	103.7107	103.7107	5.4e-12
2020-04-15	101.0526	101.0526	1.2e-12
2020-04-16	101.4107	101.4107	9.5e-13
2020-04-17	101.7589	101.7589	7.7e-13
2020-04-18	102.0976	102.0976	6.3e-13
2020-04-19	102.4272	102.4272	4.4e-13
2020-04-20	105.8359	105.8359	4.3e-12
2020-04-21	103.0597	103.0597	1.7e-13
2020-04-22	103.3633	103.3633	4.3e-14
2020-04-23	103.6588	103.6588	7.1e-14
2020-04-24	103.9463	103.9463	1.8e-13
2020-04-25	104.2263	104.2263	2.8e-13
2020-04-26	107.6397	107.6397	3.7e-12
2020-04-27	104.7645	104.7645	5.0e-13
2020-04-28	105.0231	105.0231	5.5e-13
2020-04-29	105.2751	105.2751	6.1e-13

Timing

Median of all measured repetitions.

Step	Effect Prophet	Python Prophet	Effect Prophet is
Fit	11.3 ms	18.6 ms	1.6× faster
Predict	1.06 ms	3.63 ms	3.4× faster
Save model to JSON	26 µs	481 µs	18× faster
Load model from JSON	135 µs	2.47 ms	18× faster
Load and predict in a new process	167 ms	1.42 s	8.5× 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": [],
    "regressors": [
      {
        "name": "promotion",
        "priorScale": 10,
        "standardization": "never",
        "mode": "multiplicative"
      }
    ]
  },
  "pythonOptimizer": {
    "algorithm": "Auto",
    "maxIterations": 10000,
    "newtonFallback": true,
    "sigFigs": 12
  }
}