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. Dimensions: 16382 cells, 18771 genes. 14 cell types (avg. 1170±1703 cells per cell type).
Raw Scaled
RankNameMean scoreAccuracyAccuracyF1 scoreF1 scoreMacro F1 scoreMacro F1 scoreMemory (GB)Runtime (min)CPU (%)PaperYearLibrary
1Logistic regression (log CP10k)0.960.980.980.980.980.950.942.201.38334.202013v1.1.3
2Seurat reference mapping (SCTransform)0.960.980.980.980.980.920.9220.1015.17108.402021v4.3.0
3XGBoost (log CP10k)0.930.970.970.970.970.860.852.6013.85382.802016v1.6.2
4K-neighbors classifier (log CP10k)0.910.960.950.960.950.850.832.302.90256.801967v1.1.3
5XGBoost (log scran)0.900.960.950.960.940.830.818.5021.23253.302016v1.6.2
6scANVI (Seurat v3 2000 HVG)0.890.980.970.970.970.750.735.6021.822,362.702021v0.20.0
7K-neighbors classifier (log scran)0.860.950.930.950.930.740.718.508.50120.101967v1.1.3
8scANVI (All genes)0.850.970.960.970.960.660.644.6025.002,716.102021v0.20.0
9Multilayer perceptron (log scran)0.830.930.910.940.920.670.658.5014.351,016.701990v1.1.3
10scArches+scANVI (All genes)0.810.960.950.950.940.570.545.80109.422,074.102021v0.20.0
11scArches+scANVI (Seurat v3 2000 HVG)0.810.960.950.950.930.570.534.6055.181,723.802021v0.20.0
12Multilayer perceptron (log CP10k)0.690.810.760.840.800.550.512.202.001,422.701990v1.1.3
13Logistic regression (log scran)0.650.830.790.800.750.440.398.5016.58120.502013v1.1.3
14Majority Vote0.040.340.190.17-0.040.04-0.041.500.8022.402022v0.7.0