Table 4 Comparison of predictive performances of FRF and MRF for 8 different dose points using CCLE data.

From: Functional random forest with applications in dose-response predictions

Drug

Model

Correlation

Dose 1

Dose 2

Dose 3

Dose 4

Dose 5

Dose 6

Dose 7

Dose 8

Mean

Erlotinib

MRF

0.0293

0.2014

0.1877

0.2901

0.3915

0.4813

0.4942

0.4071

0.3103

FRF

0.0662

0.1781

0.2138

0.3256

0.4378

0.4955

0.5094

0.4100

0.3296

Nilotinib

MRF

−0.0725

0.1966

0.1550

0.2860

0.3734

0.4255

0.3888

0.1830

0.2420

FRF

−0.0776

0.1360

0.2186

0.3306

0.4182

0.4546

0.4310

0.2502

0.2702

PD-0325901

MRF

0.1402

0.3722

0.4842

0.5395

0.5776

0.5871

0.5668

0.5181

0.4732

FRF

0.2013

0.4397

0.5239

0.5798

0.6067

0.6078

0.5952

0.5426

0.5121

PLX-4720

MRF

−0.0522

−0.0137

0.0885

0.1818

0.3986

0.4682

0.5018

0.3732

0.2433

FRF

−0.0045

0.1297

0.1259

0.2434

0.4028

0.4779

0.4973

0.3772

0.2812

TAE-684

MRF

0.1068

0.1485

0.0045

0.1509

0.3236

0.3448

0.2914

0.2874

0.2072

FRF

0.0978

0.1615

0.0541

0.2358

0.3654

0.3867

0.3736

0.3008

0.2470

  1. All the models are built using 150 trees, m = 10 node splitting features and minimum leaf size of 10.
Back to article page