Pancreas (random split with label noise)

Human pancreatic islet scRNA-seq data from 6 datasets across technologies (CEL-seq, CEL-seq2, Smart-seq2, inDrop, Fluidigm C1, and SMARTER-seq). Split into train/test randomly with 20% label noise.

Pancreas (random split with label noise)

Rank	F1 score	Macro F1 score	Accuracy	Memory (GB)	Runtime (min)	CPU (%)	Name	Code	Year
14	0.17	0.04	0.34	3.60	4.18	13.40	Majority Vote	v0.6.0	2022
13	0.19	0.07	0.18	3.70	6.12	10.00	Random Labels	v0.6.0	2022
7	0.93	0.82	0.93	6.50	6.08	258.60	K-neighbors classifier (log CPM)	v1.1.2	1967
1	0.97	0.84	0.97	6.50	2.27	558.90	Logistic regression (log CPM)	v1.1.2	2013
11	0.88	0.60	0.86	6.50	7.77	2,319.90	Multilayer perceptron (log CPM)	v1.1.2	1990
2	0.97	0.85	0.97	8.90	1.70	2,098.20	XGBoost (log CPM)	v1.6.2	2016
4	0.97	0.68	0.97	5.80	33.85	1,362.70	scANVI (Seurat v3 2000 HVG)	v0.17.1	2021
5	0.95	0.60	0.96	6.00	73.57	1,307.70	scANVI (All genes)	v0.17.1	2021
6	0.95	0.75	0.95	22.50	20.28	150.80	K-neighbors classifier (log scran)	v1.1.2	1967
9	0.94	0.53	0.95	6.60	78.08	1,042.10	scArches+scANVI (Seurat v3 2000 HVG)	v0.17.1	2021
12	0.80	0.47	0.82	22.50	13.50	125.40	Logistic regression (log scran)	v1.1.2	2013
8	0.94	0.53	0.95	9.70	43.55	1,473.00	scArches+scANVI (All genes)	v0.17.1	2021
3	0.96	0.89	0.96	23.10	13.97	351.00	XGBoost (log scran)	v1.6.2	2016
10	0.93	0.66	0.92	22.50	24.93	983.30	Multilayer perceptron (log scran)	v1.1.2	1990