Skip to main content

Table 2 The parameters and set of values for various off-the-shelf baselines

From: Machine learning predicts nucleosome binding modes of transcription factors

Method Tuning parameters
Logistic regression The norm used in the penalization: none, L1, L2, elastic net
Regularization coefficient: 100, 10, 1, 0.1, 0.01
Subsequence length: 3, 4, 5, 6
k-nearest neighbors Number of neighbors to use: 1, 3, 5, 7, 9, 11, 13, 15, 17, 19, 21
Contribution of members in the neighborhood: uniform, distance
Distance metric: Euclidean, Manhattan, Minkowski
Subsequence length: 3, 4, 5, 6
Support vector machine Kernel: Linear, Polynomial, RBF, Sigmoid
Regularization parameter (C): 50, 10, 1.0, 0.1, 0.01
Subsequence length: 3, 4, 5, 6
Random forest The number of features to consider when looking for the best split: sqrt(num features), log2(num features)
The number of trees in the forest: 10, 100, 200, 500, 1000
Subsequence length: 3, 4, 5, 6