Data projekt 1 – Neebile

Predikce v praxi
data projekt 1

Tento článek slouží jako appendix k sérii „Predikce v praxi“. Obsahuje pouze neupravené výsledky experimentů bez interpretace pro projekt 1. Hlavní článek série zde.

Metriky

Statistické modely

Statistic - all history
model	wape	mae	rmsle	direct_wape	trim_wape	core_wape	stable_wape	scaled_wape
AutoARIMA	166,6883	30,8894	0,8714	104,1104	94,0708	76,5071	93,5828	0,6097
SeasonalNaive	416,0951	50,9679	1,2658	347,3453	168,3804	135,8278	189,3751	0,7839
Naive	104,6668	36,6459	0,8884	51,9482	87,2913	74,7431	68,8883	0,5757
CrostonSBA	196,1718	33,911	0,9003	150,6484	99,7924	79,4747	106,4077	0,6153
TSB	203,9044	36,9866	0,9237	158,6596	106,9412	83,9276	111,1739	0,6426

Statistic - history > 24
model	wape	mae	rmsle	direct_wape	trim_wape	core_wape	stable_wape	scaled_wape
AutoARIMA	117,1419	32,5995	0,7912	56,4051	78,6357	68,8974	69,378	0,5556
SeasonalNaive	227,0287	52,7323	1,1348	160,7776	139,1377	114,5277	147,26	0,7149
AutoETS	99,7209	38,2479	0,7845	55,6208	77,819	68,4123	63,4523	0,5465
AutoCES	133,4398	34,466	0,8552	80,8756	85,8584	78,9809	87,7502	0,5866
AutoTheta	88,8216	37,1278	0,788	38,5308	73,6599	64,8364	59,6435	0,52
Naive	99,7762	39,5834	0,8644	49,6749	86,8121	74,5531	67,2045	0,5858
CrostonSBA	123,0062	35,0673	0,7949	78,3058	82,0228	69,9322	74,511	0,5486
TSB	132,085	38,5189	0,8214	87,7986	89,6233	75,5463	80,3348	0,5764

Statistic - history 12-24
model	wape	mae	rmsle	direct_wape	trim_wape	core_wape	stable_wape	scaled_wape
AutoARIMA	223,2485	31,5138	1,2909	161,814	82,7341	78,5836	142,4324	0,6464
SeasonalNaive	357,9693	41,5147	1,5175	293,9443	135,5538	135,2205	312,5703	0,8187
AutoCES	193,0068	35,4906	1,3739	121,8436	85,5886	79,7335	125,3793	0,6447
AutoTheta	128,7247	27,487	1,2098	62,1629	65,9886	66,0107	89,5323	0,592
Naive	203,8938	36,7882	1,3725	127,148	74,0838	74,4703	131,7577	0,6317
CrostonSBA	227,0647	30,4668	1,2227	171,5082	110,6259	102,2184	148,1473	0,6916
TSB	182,84	28,6746	1,1668	124,3206	84,0143	78,766	119,0055	0,6273

Statistic - history 6-11
model	wape	mae	rmsle	direct_wape	trim_wape	core_wape	stable_wape	scaled_wape
AutoARIMA	1105,3406	55,5656	2,0027	1096,3657	836,0242	457,5393	533,0963	1,7421
SeasonalNaive	3840,5021	91,4778	3,1216	3798,327	1417,7726	1057,0907	801,5135	2,2195
AutoCES	417,162	41,1355	1,4935	195,4198	304,0644	259,0544	246,4548	0,9321
AutoTheta	552,0683	39,8995	1,0407	-257,7317	226,5154	190,5365	216,2036	0,8185
Naive	284,3286	36,9506	1,2253	246,6522	169,0046	151,3901	119,0519	0,5921
CrostonSBA	1548,0592	53,6857	2,3676	1546,9739	774,9956	439,168	659,6752	1,8064
TSB	1576,3467	54,8354	2,3998	1575,7229	800,2285	433,9093	670,1465	1,9107

Statistic - history 1-5
model	wape	mae	rmsle	direct_wape	trim_wape	core_wape	stable_wape	scaled_wape
AutoARIMA	107,3593	63,4429	1,6361	9,0009	100	nan	73,9643	0,7996
SeasonalNaive	100,3237	65,2667	1,5266	18,9234	nan	nan	90,2068	0,8027
Naive	128,6416	64,0833	1,398	34,6276	nan	nan	108,7132	1,0097
CrostonSBA	111,8605	71,2567	1,682	21,818	nan	nan	83,6154	0,8513
TSB	110,1271	67,066	1,5717	27,2213	nan	nan	81,7536	0,8623

Machine learning modely

ML - all history
model	wape	mae	rmsle	direct_wape	trim_wape	core_wape	stable_wape	scaled_wape
XGBRegressor	97,9691	23,6139	0,694	60,1406	61,7398	59,2089	64,0894	0,4446
DecisionTreeRegressor	113,6364	25,2365	0,7275	63,7304	67,9523	61,4097	75,3321	0,4654
RandomForestRegressor	106,9287	24,2712	0,7127	75,0828	64,2291	60,6401	67,5486	0,4592
GradientBoostingRegressor	138,3042	26,8507	0,7947	105,3901	76,451	68,6938	78,8409	0,5318
ExtraTreesRegressor	121,3819	24,8232	0,7466	91,3831	69,0022	63,6157	71,363	0,4872
Ridge	261,7974	33,8969	0,9114	230,6663	112,1865	93,7249	137,996	0,6872
Lasso	262,0528	33,8985	0,9114	230,8806	112,2361	93,7518	138,1298	0,6871
ElasticNet	261,9575	33,8922	0,9114	230,7453	112,1997	93,7192	138,078	0,6869
QuantileRegressor	207,3545	31,0306	0,878	167,9003	94,1082	80,4029	111,7842	0,6259

ML - history > 24
model	wape	mae	rmsle	direct_wape	trim_wape	core_wape	stable_wape	scaled_wape
XGBRegressor	73,9811	21,7326	0,6381	37,5239	57,5819	53,2593	48,8667	0,4291
DecisionTreeRegressor	77,1131	22,8838	0,6648	42,1487	61,7678	54,6547	49,8018	0,4402
RandomForestRegressor	73,9434	21,9198	0,6418	41,4268	56,8637	52,799	48,9135	0,4286
GradientBoostingRegressor	83,2231	24,1617	0,6912	51,3167	64,9539	58,5764	53,9422	0,4706
ExtraTreesRegressor	76,9409	22,1839	0,6544	46,6784	59,2338	54,3684	49,8236	0,4398
Ridge	132,1849	30,7511	0,7833	100,8604	90,2069	75,2105	77,3003	0,5782
Lasso	132,2429	30,7672	0,7833	100,8582	90,2132	75,2015	77,3284	0,5781
ElasticNet	132,2142	30,7658	0,7833	100,7962	90,1912	75,1831	77,3153	0,5781
QuantileRegressor	108,5282	28,2255	0,7628	68,6993	76,8018	65,7387	66,0736	0,5266

ML - history 12-24
model	wape	mae	rmsle	direct_wape	trim_wape	core_wape	stable_wape	scaled_wape
XGBRegressor	146,7095	35,2481	0,8669	108,48	66,4416	104,5445	91,42	0,6632
DecisionTreeRegressor	249,634	28,6607	0,8523	208,4736	79,1087	85,5241	183,8219	0,6311
RandomForestRegressor	244,6927	26,2442	0,8051	205,1327	77,1677	78,8656	157,1895	0,6241
GradientBoostingRegressor	268,5272	27,6209	0,8321	229,7479	71,6253	87,8368	185,0579	0,6563
ExtraTreesRegressor	226,1194	25,9269	0,7958	183,9585	75,037	75,6764	144,1695	0,6092
Ridge	205,5025	28,8271	0,8536	157,9195	82,7296	76,0827	131,0599	0,6308
Lasso	205,5394	28,97	0,8551	157,8146	83,1335	76,5073	131,2171	0,6339
ElasticNet	205,0256	28,9374	0,8544	157,3392	82,8628	76,2101	130,8498	0,6324
QuantileRegressor	133,9218	31,0026	0,8586	66,8838	61,2438	61,2148	91,0335	0,5513

Statistic history 6-11
model	wape	mae	rmsle	direct_wape	trim_wape	core_wape	stable_wape	scaled_wape
XGBRegressor	150.3444	14.9619	1.9675	-2.444	109.3206	102.6793	85.6634	0.4178
DecisionTreeRegressor	1499.1687	26.2399	2.3373	1367.916	113.4239	402.9385	176.6635	1.1297
RandomForestRegressor	1498.9482	26.2624	2.3386	1364.1214	113.0966	403.0532	176.8506	1.1313
GradientBoostingRegressor	1498.3351	26.2233	2.3368	1367.5609	113.4677	402.7285	176.575	1.1294
ExtraTreesRegressor	2045.6269	31.6264	2.6239	2043.1151	553.5524	478.0101	556.077	1.8487
Ridge	3867.2315	60.1437	03.09	3867.2315	1082.2986	889.4863	1766.1144	2.7825
Lasso	3866.7816	60.1366	3.0899	3866.7816	1082.1767	889.3848	1765.8864	2.7824
ElasticNet	3867.2393	60.1438	03.09	3867.2393	1082.3007	889.4881	1766.1184	2.7825
QuantileRegressor	4672.4	72.8118	3.2489	4672.4	1298.9879	1071.1127	2192.8896	2.8882

ML - history 1-5
model	wape	mae	rmsle	direct_wape	trim_wape	core_wape	stable_wape	scaled_wape
XGBRegressor	183,7108	63,7049	1,5819	113,8934	153,7352	183,7108	132,9614	1,0103
DecisionTreeRegressor	155,4078	68,8532	1,6217	76,2339	128,6007	155,4078	115,4768	0,9673
RandomForestRegressor	153,4037	68,4205	1,6199	73,257	126,781	153,4037	114,3456	0,966
GradientBoostingRegressor	153,0601	68,3464	1,6197	72,7465	126,469	153,0601	114,1517	0,9658
ExtraTreesRegressor	154,5453	68,667	1,6209	74,9527	127,8175	154,5453	114,9899	0,9667
Ridge	160,0767	59,0076	1,5402	82,7292	131,1265	160,0767	118,228	0,9745
Lasso	182,1459	59,6237	1,5302	111,3923	155,0336	182,1459	132,276	1,0061
ElasticNet	182,0096	59,6243	1,5303	111,2152	154,8668	182,0096	132,1893	1,0059
QuantileRegressor	141,8712	55,3082	1,4962	54,8299	122,8748	141,8712	108,6306	0,9586

Deep learning modely

DL - all history
model	wape	mae	rmsle	direct_wape	trim_wape	core_wape	stable_wape	scaled_wape
TFT	97,3795	26,1338	0,7964	52,8239	59,4895	61,1367	61,8653	0,4881
DeepAR	84,5275	23,9573	0,7842	30,6076	52,6275	55,4608	57,7031	0,4423
NHiTS	87,4824	27,2496	0,7548	49,5937	65,0793	67,8852	57,2459	0,4761
PatchTST	116,2095	24,147	0,7319	76,876	62,2834	59,4683	65,8495	0,4779

DL - history > 24
model	wape	mae	rmsle	direct_wape	trim_wape	core_wape	stable_wape	scaled_wape
TFT	70,8446	26,2639	0,7184	20,5002	52,3412	54,8432	51,6678	0,4452
DeepAR	58,2096	22,8979	0,6618	16,4755	45,9445	51,1129	43,0649	0,386
NHiTS	62,696	27,1243	0,6803	23,8414	53,6608	56,3535	44,3809	0,4244
PatchTST	58,9393	24,8246	0,6376	12,2064	48,9178	51,2714	41,7765	0,3895

DL - history 12-24
model	wape	mae	rmsle	direct_wape	trim_wape	core_wape	stable_wape	scaled_wape
TFT	96.2817	30.3649	0.7853	6.1694	86.1307	96.2817	70.3828	0.5986
DeepAR	99.6592	34.3726	1.2565	-16.8609	99.4292	99.6592	84.9898	0.7032
NHiTS	87.6493	34.975	0.9572	-24.2647	80.7647	87.6493	76.5592	0.6441
PatchTST	315.059	46.683	0.7871	233.543	309.364	315.059	202.709	0.6061

DL - history 6-11
model	wape	mae	rmsle	direct_wape	trim_wape	core_wape	stable_wape	scaled_wape
TFT	89.3762	25.2888	1.1813	-30.2471	82.5929	78.6624	79.7236	0.3974
DeepAR	499.683	36.531	1.8086	455.526	152.672	229.605	193.527	1.1196
NHiTS	129.517	27.213	1.6788	-37.2184	100	101.571	97.4232	0.5145
PatchTST	873.137	40.1397	2.1333	888.734	313.832	309.227	324.523	1.5721

DL - history 1-5
model	wape	mae	rmsle	direct_wape	trim_wape	core_wape	stable_wape	scaled_wape
TFT	94.3054	44.2639	2.6168	-93.3834	99.9134	97.513	95.2057	0.7962
DeepAR	73.8493	34.3997	1.1856	-20.5986	66.7327	71.4719	61.9897	0.5734
NHiTS	402.093	153.43	2.7801	295.849	197.017	429.556	238.544	1.204
PatchTST	543.109	147.434	2.3157	500.831	534.465	493.17	399.904	2.2026

Subscribe for Latest News

Better forecasting thanks to AI

Business Info.

IČO: 172 28 018

DIČ: CZ 172 28 018

Data Box ID: ykwdnxf

sales@neebile.cz

Jičínská 226/17, Praha, Žižkov, PSČ 130 00 Česká republika

(910) 658-2992

© 2025 Vytvořeno DigitalWays