Autoencoder-kNN meta-model based data characterization approach for an automated selection of AI algorithms

Garouani, Moncef; Ahmad, Adeel; Bouneffa, Mourad; Hamlich, Mohamed

doi:10.1186/s40537-023-00687-7

Journal of Big Data

Table 8 Results of the RF, XGB, KNN, AeKNN meta-models for recommending optimal pipelines for test data

From: Autoencoder-kNN meta-model based data characterization approach for an automated selection of AI algorithms

Dataset	AeKNN			KNN			XGB			RF
Dataset	Acc	F1-score	AUC	Acc	F1-score	AUC	Acc	F1-score	AUC	Acc	F1-score	AUC
APSFailure	0.9921	0.9823	0.9191	0.991	0.9105	0.7150	0.9673	0.8868	0.9468	0.8950	0.6920	0.6920
Higgs	0.7283	0.8743	0.7283	0.7260	0.5725	0.4865	0.6801	0.5631	0.4771	0.6072	0.5867	0.5867
CustSat	0.8155	0.9250	0.9654	0.7998	0.7558	0.7063	0.8715	0.5355	0.6320	0.7382	0.7542	0.7542
car	0.9999	0.9635	0.7724	0.9754	0.7124	0.9608	0.9462	0.6467	0.8162	0.8549	0.6884	0.6884
kr-vs-kp	0.9976	0.9246	0.7631	0.9209	0.7309	0.6449	0.7593	0.4963	0.7765	0.6532	0.4867	0.4867
airlines	0.6982	0.5887	0.8627	0.6758	0.5953	0.5823	0.7094	0.6289	0.5429	0.5927	0.3532	0.3532
vehicle	0.888	0.8515	0.9610	0.8415	0.7610	0.7480	0.9027	0.6762	0.8822	0.6591	0.6386	0.6386
MiniBooNE	0.9645	0.9715	0.8550	0.9423	0.7888	0.7393	0.8903	0.8098	0.6143	0.8343	0.5583	0.5583
jannis	0.7229	0.7229	0.7338	0.6719	0.5549	0.6879	0.6845	0.4215	0.4815	0.6171	0.4506	0.4506
nomao	0.9708	0.9343	0.8007	0.9570	0.6940	0.8594	0.7987	0.6817	0.5227	0.6995	0.4965	0.4965
Credi-g	0.7921	0.9381	0.9381	0.7188	0.4923	0.6983	0.5739	0.5299	0.5899	0.6121	0.5916	0.5916
Kc1	0.8793	0.9321	0.7333	0.8552	0.6287	0.6887	0.7697	0.4337	0.7127	0.7097	0.6892	0.6892
Cnae-9	0.9671	0.7998	0.8941	0.8803	0.8962	0.7868	0.8365	0.6830	0.8525	0.7922	0.5892	0.5892
albert	0.8759	0.8394	0.9124	0.8005	0.7565	0.6340	0.8288	0.6753	0.5528	0.7981	0.7046	0.7046
Numerai28.6	0.5207	0.3747	0.6302	0.4433	0.1803	0.4228	0.4836	0.2936	0.2076	0.4229	0.2971	0.3571
segment	0.9735	0.9130	0.8900	0.9681	0.7416	0.7651	0.9542	0.7642	0.8242	0.9337	0.8767	0.8767
Covertype	0.8344	0.6886	0.7204	0.7307	0.6502	0.6007	0.7890	0.4530	0.7592	0.6521	0.4126	0.4126
KDDCup	0.9740	0.9571	0.9660	0.9500	0.8330	0.9876	0.9331	0.7796	0.9491	0.8934	0.7634	0.7634
shuttle	0.9362	0.9653	0.9727	0.9905	0.9100	0.9700	0.9649	0.6289	0.6889	0.8429	0.6764	0.6764
Gas_Sens-uci	0.9843	0.6161	0.8748	0.9739	0.8569	0.6979	0.9468	0.8298	0.6708	0.9256	0.7591	0.7591

The results for all meta-models are presented jointly, and the best ones are highlighted in bold

Back to article page