Three automatic changepoints, minmax scaling

match ← All logistic growth benchmarks

96 daily rows rising toward a ceiling of about 100, with a little noise. Prediction covers the history and 24 more days. Changepoints are spread over the first 80% of the history. Values are scaled between the lowest floor and the highest ceiling.

Input

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

Output

Largest difference: 2.4e-9 (< 0.0001% of the input's range).

Every forecast row from both libraries.

Date	Effect Prophet	Python Prophet	Difference
2020-01-01	16.8952	16.8952	8.5e-11
2020-01-02	17.1318	17.1318	8.7e-11
2020-01-03	17.3885	17.3885	8.8e-11
2020-01-04	17.6669	17.6669	8.9e-11
2020-01-05	17.9687	17.9687	8.9e-11
2020-01-06	18.2957	18.2957	8.8e-11
2020-01-07	18.6496	18.6496	8.7e-11
2020-01-08	19.0326	19.0326	8.4e-11
2020-01-09	19.4465	19.4465	8.1e-11
2020-01-10	19.8937	19.8937	7.7e-11
2020-01-11	20.3763	20.3763	7.1e-11
2020-01-12	20.8966	20.8966	6.4e-11
2020-01-13	21.4570	21.4570	5.5e-11
2020-01-14	22.0599	22.0599	4.5e-11
2020-01-15	22.7078	22.7078	3.2e-11
2020-01-16	23.4032	23.4032	1.8e-11
2020-01-17	24.1485	24.1485	8.2e-13
2020-01-18	24.9462	24.9462	1.8e-11
2020-01-19	25.7985	25.7985	4.0e-11
2020-01-20	26.7078	26.7078	6.5e-11
2020-01-21	27.6761	27.6761	9.2e-11
2020-01-22	28.7053	28.7053	1.2e-10
2020-01-23	29.7970	29.7970	1.6e-10
2020-01-24	30.9526	30.9526	1.9e-10
2020-01-25	32.1731	32.1731	2.3e-10
2020-01-26	33.4591	33.4591	2.8e-10
2020-01-27	34.2828	34.2828	2.5e-10
2020-01-28	35.1314	35.1314	2.2e-10
2020-01-29	36.0049	36.0049	1.9e-10
2020-01-30	36.9031	36.9031	1.6e-10
2020-01-31	37.8257	37.8257	1.2e-10
2020-02-01	38.7723	38.7723	8.4e-11
2020-02-02	39.7426	39.7426	4.5e-11
2020-02-03	40.7360	40.7360	3.8e-12
2020-02-04	41.7519	41.7519	3.9e-11
2020-02-05	42.7896	42.7896	8.4e-11
2020-02-06	43.8483	43.8483	1.3e-10
2020-02-07	44.9271	44.9271	1.8e-10
2020-02-08	46.0251	46.0251	2.3e-10
2020-02-09	47.1413	47.1413	2.8e-10
2020-02-10	48.2744	48.2744	3.3e-10
2020-02-11	49.4232	49.4232	3.8e-10
2020-02-12	50.5865	50.5865	4.4e-10
2020-02-13	51.7629	51.7629	4.9e-10
2020-02-14	52.9510	52.9510	5.4e-10
2020-02-15	54.1493	54.1493	6.0e-10
2020-02-16	55.3562	55.3562	6.5e-10
2020-02-17	56.5702	56.5702	7.1e-10
2020-02-18	57.7896	57.7896	7.6e-10
2020-02-19	59.0129	59.0129	8.1e-10
2020-02-20	60.2384	60.2384	8.6e-10
2020-02-21	61.2853	61.2853	7.5e-10
2020-02-22	62.3318	62.3318	6.4e-10
2020-02-23	63.3769	63.3769	5.2e-10
2020-02-24	64.4197	64.4197	4.1e-10
2020-02-25	65.4591	65.4591	3.0e-10
2020-02-26	66.4943	66.4943	1.9e-10
2020-02-27	67.5242	67.5242	7.5e-11
2020-02-28	68.5479	68.5479	3.4e-11
2020-02-29	69.5645	69.5645	1.4e-10
2020-03-01	70.5732	70.5732	2.5e-10
2020-03-02	71.5732	71.5732	3.5e-10
2020-03-03	72.5636	72.5636	4.5e-10
2020-03-04	73.5436	73.5436	5.5e-10
2020-03-05	74.5126	74.5126	6.4e-10
2020-03-06	75.4699	75.4699	7.3e-10
2020-03-07	76.4149	76.4149	8.2e-10
2020-03-08	77.3469	77.3469	9.0e-10
2020-03-09	78.2654	78.2654	9.9e-10
2020-03-10	79.1699	79.1699	1.1e-9
2020-03-11	80.0599	80.0599	1.1e-9
2020-03-12	80.9351	80.9351	1.2e-9
2020-03-13	81.7950	81.7950	1.3e-9
2020-03-14	82.6394	82.6394	1.3e-9
2020-03-15	83.4680	83.4680	1.4e-9
2020-03-16	84.2805	84.2805	1.4e-9
2020-03-17	85.0768	85.0768	1.2e-9
2020-03-18	85.8566	85.8566	9.0e-10
2020-03-19	86.6199	86.6199	6.4e-10
2020-03-20	87.3667	87.3667	4.0e-10
2020-03-21	88.0968	88.0968	1.7e-10
2020-03-22	88.8103	88.8103	4.7e-11
2020-03-23	89.5071	89.5071	2.5e-10
2020-03-24	90.1875	90.1875	4.5e-10
2020-03-25	90.8513	90.8513	6.3e-10
2020-03-26	91.4989	91.4989	8.1e-10
2020-03-27	92.1302	92.1302	9.7e-10
2020-03-28	92.7455	92.7455	1.1e-9
2020-03-29	93.3450	93.3450	1.3e-9
2020-03-30	93.9288	93.9288	1.4e-9
2020-03-31	94.4972	94.4972	1.5e-9
2020-04-01	95.0505	95.0505	1.6e-9
2020-04-02	95.5888	95.5888	1.7e-9
2020-04-03	96.1125	96.1125	1.8e-9
2020-04-04	96.6218	96.6218	1.9e-9
2020-04-05	97.1170	97.1170	2.0e-9
2020-04-06	97.5984	97.5984	2.1e-9
2020-04-07	98.0663	98.0663	2.1e-9
2020-04-08	98.5210	98.5210	2.2e-9
2020-04-09	98.9629	98.9629	2.2e-9
2020-04-10	99.3923	99.3923	2.3e-9
2020-04-11	99.8093	99.8093	2.3e-9
2020-04-12	100.2145	100.2145	2.3e-9
2020-04-13	100.6081	100.6081	2.3e-9
2020-04-14	100.9904	100.9904	2.4e-9
2020-04-15	101.3618	101.3618	2.4e-9
2020-04-16	101.7225	101.7225	2.4e-9
2020-04-17	102.0729	102.0729	2.4e-9
2020-04-18	102.4133	102.4133	2.4e-9
2020-04-19	102.7440	102.7440	2.4e-9
2020-04-20	103.0653	103.0653	2.4e-9
2020-04-21	103.3775	103.3775	2.3e-9
2020-04-22	103.6808	103.6808	2.3e-9
2020-04-23	103.9756	103.9756	2.3e-9
2020-04-24	104.2622	104.2622	2.3e-9
2020-04-25	104.5408	104.5408	2.3e-9
2020-04-26	104.8117	104.8117	2.2e-9
2020-04-27	105.0752	105.0752	2.2e-9
2020-04-28	105.3315	105.3315	2.2e-9
2020-04-29	105.5809	105.5809	2.2e-9

Timing

Median of all measured repetitions.

Step	Effect Prophet	Python Prophet	Effect Prophet is
Fit	9.52 ms	18.0 ms	1.9× faster
Predict	778 µs	2.91 ms	3.7× faster
Save model to JSON	22 µs	453 µs	21× faster
Load model from JSON	145 µs	2.31 ms	16× faster
Load and predict in a new process	154 ms	1.50 s	9.7× faster

Settings

The case declaration both libraries ran.

{
  "kind": "map",
  "comparison": {
    "kind": "equivalent-objective",
    "evidenceId": "logistic-python-reconciliation-v1"
  },
  "configuration": {
    "growth": "logistic",
    "scaling": "minmax",
    "seasonalityMode": "additive",
    "holidaysMode": "additive",
    "changepoints": {
      "mode": "auto",
      "count": 3,
      "range": 0.8
    },
    "changepointPriorScale": 0.05,
    "seasonalities": [],
    "events": [],
    "regressors": []
  },
  "pythonOptimizer": {
    "algorithm": "Auto",
    "maxIterations": 10000,
    "newtonFallback": true,
    "sigFigs": 12
  }
}