99 rows: just under the optimizer switch

match ← All logistic growth benchmarks

One row short of 100, so both libraries fit with Newton's method.

Input

benchmark/inputs/v2/logistic-basic-99.json: 99 history rows and 123 forecast rows.

Output

Largest difference: 3.8e-11 (< 0.0001% of the input's range).

Every forecast row from both libraries.

Date	Effect Prophet	Python Prophet	Difference
2020-01-01	14.1870	14.1870	2.5e-11
2020-01-02	14.7029	14.7029	2.6e-11
2020-01-03	15.2344	15.2344	2.6e-11
2020-01-04	15.7817	15.7817	2.7e-11
2020-01-05	16.3450	16.3450	2.7e-11
2020-01-06	16.9246	16.9246	2.8e-11
2020-01-07	17.5207	17.5207	2.8e-11
2020-01-08	18.1333	18.1333	2.9e-11
2020-01-09	18.7628	18.7628	3.0e-11
2020-01-10	19.4092	19.4092	3.0e-11
2020-01-11	20.0727	20.0727	3.1e-11
2020-01-12	20.7533	20.7533	3.1e-11
2020-01-13	21.4511	21.4511	3.2e-11
2020-01-14	22.1662	22.1662	3.2e-11
2020-01-15	22.8986	22.8986	3.3e-11
2020-01-16	23.6482	23.6482	3.3e-11
2020-01-17	24.4152	24.4152	3.4e-11
2020-01-18	25.1993	25.1993	3.4e-11
2020-01-19	26.0004	26.0004	3.4e-11
2020-01-20	26.8185	26.8185	3.5e-11
2020-01-21	27.6534	27.6534	3.5e-11
2020-01-22	28.5049	28.5049	3.6e-11
2020-01-23	29.3726	29.3726	3.6e-11
2020-01-24	30.2564	30.2564	3.6e-11
2020-01-25	31.1560	31.1560	3.6e-11
2020-01-26	32.0709	32.0709	3.7e-11
2020-01-27	33.0009	33.0009	3.7e-11
2020-01-28	33.9454	33.9454	3.7e-11
2020-01-29	34.9040	34.9040	3.7e-11
2020-01-30	35.8762	35.8762	3.7e-11
2020-01-31	36.8615	36.8615	3.8e-11
2020-02-01	37.8593	37.8593	3.8e-11
2020-02-02	38.8690	38.8690	3.8e-11
2020-02-03	39.8900	39.8900	3.8e-11
2020-02-04	40.9216	40.9216	3.8e-11
2020-02-05	41.9631	41.9631	3.8e-11
2020-02-06	43.0137	43.0137	3.8e-11
2020-02-07	44.0728	44.0728	3.7e-11
2020-02-08	45.1395	45.1395	3.7e-11
2020-02-09	46.2131	46.2131	3.7e-11
2020-02-10	47.2926	47.2926	3.7e-11
2020-02-11	48.3774	48.3774	3.7e-11
2020-02-12	49.4665	49.4665	3.6e-11
2020-02-13	50.5591	50.5591	3.6e-11
2020-02-14	51.6543	51.6543	3.6e-11
2020-02-15	52.7512	52.7512	3.5e-11
2020-02-16	53.8489	53.8489	3.5e-11
2020-02-17	54.9466	54.9466	3.4e-11
2020-02-18	56.0434	56.0434	3.4e-11
2020-02-19	57.1384	57.1384	3.3e-11
2020-02-20	58.2308	58.2308	3.3e-11
2020-02-21	59.3196	59.3196	3.2e-11
2020-02-22	60.4041	60.4041	3.2e-11
2020-02-23	61.4834	61.4834	3.1e-11
2020-02-24	62.5568	62.5568	3.1e-11
2020-02-25	63.6234	63.6234	3.0e-11
2020-02-26	64.6825	64.6825	2.9e-11
2020-02-27	65.7334	65.7334	2.9e-11
2020-02-28	66.7753	66.7753	2.8e-11
2020-02-29	67.8077	67.8077	2.7e-11
2020-03-01	68.8297	68.8297	2.7e-11
2020-03-02	69.8410	69.8410	2.6e-11
2020-03-03	70.8407	70.8407	2.5e-11
2020-03-04	71.8285	71.8285	2.5e-11
2020-03-05	72.8038	72.8038	2.4e-11
2020-03-06	73.7661	73.7661	2.3e-11
2020-03-07	74.7151	74.7151	2.2e-11
2020-03-08	75.6502	75.6502	2.2e-11
2020-03-09	76.5711	76.5711	2.1e-11
2020-03-10	77.4776	77.4776	2.0e-11
2020-03-11	78.3693	78.3693	2.0e-11
2020-03-12	79.2459	79.2459	1.9e-11
2020-03-13	80.1072	80.1072	1.8e-11
2020-03-14	80.9531	80.9531	1.8e-11
2020-03-15	81.7834	81.7834	1.7e-11
2020-03-16	82.5979	82.5979	1.6e-11
2020-03-17	83.3965	83.3965	1.6e-11
2020-03-18	84.1793	84.1793	1.5e-11
2020-03-19	84.9461	84.9461	1.4e-11
2020-03-20	85.6970	85.6970	1.4e-11
2020-03-21	86.4320	86.4320	1.3e-11
2020-03-22	87.1511	87.1511	1.3e-11
2020-03-23	87.8544	87.8544	1.2e-11
2020-03-24	88.5419	88.5419	1.2e-11
2020-03-25	89.2139	89.2139	1.1e-11
2020-03-26	89.8703	89.8703	1.1e-11
2020-03-27	90.5114	90.5114	1.0e-11
2020-03-28	91.1374	91.1374	9.5e-12
2020-03-29	91.7483	91.7483	9.1e-12
2020-03-30	92.3445	92.3445	8.6e-12
2020-03-31	92.9261	92.9261	8.2e-12
2020-04-01	93.4933	93.4933	7.8e-12
2020-04-02	94.0464	94.0464	7.4e-12
2020-04-03	94.5855	94.5855	7.0e-12
2020-04-04	95.1110	95.1110	6.6e-12
2020-04-05	95.6232	95.6232	6.2e-12
2020-04-06	96.1222	96.1222	5.9e-12
2020-04-07	96.6083	96.6083	5.6e-12
2020-04-08	97.0818	97.0818	5.2e-12
2020-04-09	97.5430	97.5430	4.9e-12
2020-04-10	97.9921	97.9921	4.6e-12
2020-04-11	98.4295	98.4295	4.3e-12
2020-04-12	98.8553	98.8553	4.1e-12
2020-04-13	99.2700	99.2700	3.8e-12
2020-04-14	99.6737	99.6737	3.5e-12
2020-04-15	100.0667	100.0667	3.3e-12
2020-04-16	100.4494	100.4494	3.1e-12
2020-04-17	100.8220	100.8220	2.9e-12
2020-04-18	101.1847	101.1847	2.7e-12
2020-04-19	101.5379	101.5379	2.5e-12
2020-04-20	101.8818	101.8818	2.3e-12
2020-04-21	102.2166	102.2166	2.1e-12
2020-04-22	102.5427	102.5427	1.9e-12
2020-04-23	102.8603	102.8603	1.8e-12
2020-04-24	103.1696	103.1696	1.6e-12
2020-04-25	103.4709	103.4709	1.5e-12
2020-04-26	103.7644	103.7644	1.4e-12
2020-04-27	104.0504	104.0504	1.2e-12
2020-04-28	104.3291	104.3291	1.1e-12
2020-04-29	104.6007	104.6007	1.0e-12
2020-04-30	104.8655	104.8655	9.0e-13
2020-05-01	105.1236	105.1236	7.8e-13
2020-05-02	105.3754	105.3754	7.1e-13

Timing

Median of all measured repetitions.

Step	Effect Prophet	Python Prophet	Effect Prophet is
Fit	16.9 ms	26.5 ms	1.6× faster
Predict	851 µs	3.25 ms	3.8× faster
Save model to JSON	25 µs	494 µs	20× faster
Load model from JSON	129 µs	2.36 ms	18× faster
Load and predict in a new process	158 ms	1.59 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": [
        "2020-02-08T00:00:00.000Z"
      ]
    },
    "changepointPriorScale": 0.05,
    "seasonalities": [],
    "events": [],
    "regressors": []
  },
  "pythonOptimizer": {
    "algorithm": "Auto",
    "maxIterations": 10000,
    "newtonFallback": true,
    "sigFigs": 12
  }
}