Values outside the bounds

match ← All logistic growth benchmarks

Noisy targets that go below zero and above the ceiling.

Input

benchmark/inputs/v2/logistic-out-of-bounds-96.json: 96 history rows and 120 forecast rows.

Output

Largest difference: 4.3e-8 (< 0.0001% of the input's range).

Every forecast row from both libraries.

Date	Effect Prophet	Python Prophet	Difference
2020-01-01	12.7277	12.7277	3.8e-8
2020-01-02	13.2447	13.2447	3.9e-8
2020-01-03	13.7794	13.7794	3.9e-8
2020-01-04	14.3324	14.3324	3.9e-8
2020-01-05	14.9038	14.9038	4.0e-8
2020-01-06	15.4941	15.4941	4.0e-8
2020-01-07	16.1035	16.1035	4.0e-8
2020-01-08	16.7324	16.7324	4.0e-8
2020-01-09	17.3809	17.3809	4.0e-8
2020-01-10	18.0494	18.0494	4.0e-8
2020-01-11	18.7381	18.7381	4.0e-8
2020-01-12	19.4470	19.4470	4.0e-8
2020-01-13	20.1765	20.1765	4.0e-8
2020-01-14	20.9267	20.9267	3.9e-8
2020-01-15	21.6975	21.6975	3.9e-8
2020-01-16	22.4891	22.4891	3.9e-8
2020-01-17	23.3014	23.3014	3.8e-8
2020-01-18	24.1345	24.1345	3.8e-8
2020-01-19	24.9883	24.9883	3.7e-8
2020-01-20	25.8626	25.8626	3.6e-8
2020-01-21	26.7572	26.7572	3.5e-8
2020-01-22	27.6720	27.6720	3.4e-8
2020-01-23	28.6067	28.6067	3.3e-8
2020-01-24	29.5608	29.5608	3.2e-8
2020-01-25	30.5342	30.5342	3.1e-8
2020-01-26	31.5263	31.5263	3.0e-8
2020-01-27	32.5366	32.5366	2.9e-8
2020-01-28	33.5646	33.5646	2.7e-8
2020-01-29	34.6098	34.6098	2.6e-8
2020-01-30	35.6714	35.6714	2.4e-8
2020-01-31	36.7488	36.7488	2.3e-8
2020-02-01	37.8411	37.8411	2.1e-8
2020-02-02	38.9477	38.9477	1.9e-8
2020-02-03	40.0677	40.0677	1.7e-8
2020-02-04	41.2001	41.2001	1.5e-8
2020-02-05	42.3441	42.3441	1.4e-8
2020-02-06	43.4986	43.4986	1.2e-8
2020-02-07	44.6626	44.6626	9.5e-9
2020-02-08	45.8352	45.8352	7.5e-9
2020-02-09	47.0151	47.0151	5.5e-9
2020-02-10	48.2014	48.2014	3.5e-9
2020-02-11	49.3929	49.3929	1.5e-9
2020-02-12	50.5883	50.5883	5.7e-10
2020-02-13	51.7867	51.7867	2.6e-9
2020-02-14	52.9868	52.9868	4.7e-9
2020-02-15	54.1875	54.1875	6.7e-9
2020-02-16	55.3876	55.3876	8.7e-9
2020-02-17	56.5859	56.5859	1.1e-8
2020-02-18	57.7813	57.7813	1.3e-8
2020-02-19	58.9726	58.9726	1.5e-8
2020-02-20	60.1588	60.1588	1.7e-8
2020-02-21	61.3387	61.3387	1.8e-8
2020-02-22	62.5114	62.5114	2.0e-8
2020-02-23	63.6757	63.6757	2.2e-8
2020-02-24	64.8306	64.8306	2.4e-8
2020-02-25	65.9753	65.9753	2.5e-8
2020-02-26	67.1088	67.1088	2.7e-8
2020-02-27	68.2301	68.2301	2.9e-8
2020-02-28	69.3386	69.3386	3.0e-8
2020-02-29	70.4334	70.4334	3.1e-8
2020-03-01	71.5138	71.5138	3.3e-8
2020-03-02	72.5790	72.5790	3.4e-8
2020-03-03	73.6286	73.6286	3.5e-8
2020-03-04	74.6619	74.6619	3.6e-8
2020-03-05	75.6783	75.6783	3.7e-8
2020-03-06	76.6774	76.6774	3.8e-8
2020-03-07	77.6589	77.6589	3.9e-8
2020-03-08	78.6223	78.6223	4.0e-8
2020-03-09	79.5672	79.5672	4.0e-8
2020-03-10	80.4936	80.4936	4.1e-8
2020-03-11	81.4010	81.4010	4.2e-8
2020-03-12	82.2895	82.2895	4.2e-8
2020-03-13	83.1587	83.1587	4.2e-8
2020-03-14	84.0088	84.0088	4.3e-8
2020-03-15	84.8395	84.8395	4.3e-8
2020-03-16	85.6509	85.6509	4.3e-8
2020-03-17	86.4431	86.4431	4.3e-8
2020-03-18	87.2162	87.2162	4.3e-8
2020-03-19	87.9701	87.9701	4.3e-8
2020-03-20	88.7052	88.7052	4.3e-8
2020-03-21	89.4215	89.4215	4.3e-8
2020-03-22	90.1192	90.1192	4.3e-8
2020-03-23	90.7986	90.7986	4.3e-8
2020-03-24	91.4598	91.4599	4.2e-8
2020-03-25	92.1033	92.1033	4.2e-8
2020-03-26	92.7292	92.7292	4.2e-8
2020-03-27	93.3378	93.3378	4.1e-8
2020-03-28	93.9294	93.9294	4.1e-8
2020-03-29	94.5044	94.5044	4.0e-8
2020-03-30	95.0632	95.0632	4.0e-8
2020-03-31	95.6060	95.6060	3.9e-8
2020-04-01	96.1332	96.1332	3.9e-8
2020-04-02	96.6451	96.6451	3.8e-8
2020-04-03	97.1422	97.1422	3.8e-8
2020-04-04	97.6248	97.6248	3.7e-8
2020-04-05	98.0932	98.0932	3.6e-8
2020-04-06	98.5479	98.5479	3.6e-8
2020-04-07	98.9892	98.9892	3.5e-8
2020-04-08	99.4174	99.4174	3.4e-8
2020-04-09	99.8330	99.8330	3.4e-8
2020-04-10	100.2363	100.2363	3.3e-8
2020-04-11	100.6277	100.6277	3.2e-8
2020-04-12	101.0076	101.0076	3.2e-8
2020-04-13	101.3762	101.3762	3.1e-8
2020-04-14	101.7340	101.7340	3.0e-8
2020-04-15	102.0812	102.0812	3.0e-8
2020-04-16	102.4183	102.4183	2.9e-8
2020-04-17	102.7456	102.7456	2.8e-8
2020-04-18	103.0633	103.0633	2.8e-8
2020-04-19	103.3718	103.3718	2.7e-8
2020-04-20	103.6715	103.6715	2.6e-8
2020-04-21	103.9626	103.9626	2.6e-8
2020-04-22	104.2454	104.2454	2.5e-8
2020-04-23	104.5202	104.5202	2.4e-8
2020-04-24	104.7873	104.7873	2.4e-8
2020-04-25	105.0470	105.0470	2.3e-8
2020-04-26	105.2995	105.2995	2.2e-8
2020-04-27	105.5452	105.5452	2.2e-8
2020-04-28	105.7842	105.7842	2.1e-8
2020-04-29	106.0168	106.0168	2.1e-8

Timing

Median of all measured repetitions.

Step	Effect Prophet	Python Prophet	Effect Prophet is
Fit	9.04 ms	16.6 ms	1.8× faster
Predict	799 µs	3.01 ms	3.8× faster
Save model to JSON	30 µs	470 µs	16× faster
Load model from JSON	149 µs	2.26 ms	15× faster
Load and predict in a new process	156 ms	1.46 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": [],
    "regressors": []
  },
  "pythonOptimizer": {
    "algorithm": "Auto",
    "maxIterations": 10000,
    "newtonFallback": true,
    "sigFigs": 12
  }
}