Regarding the K-means performance problem highlighted in Section 3.3, we randomly chose a driver and extracted driving data from cautious and reckless behavior from the training dataset and computed the clusters' centroids.
Then, we used each of the prediction models to predict the issue fix time for the test dataset and computed the performance metrics (see Sect. 4.2) for each model.
More specifically, we repeated the experiments with 10 different pairs of training and testing subsets that were randomly chosen from the whole dataset with an equal size (i.e., 31,923 subjects for our dataset) and computed the mean and standard deviation of the obtained MAEs.
We then created a bootstrap sample by randomly sampling the blocks in the dataset, and computed the bootstrap replicates of the relevant summary statistics of the expression levels.
We extracted the 353 discriminative "clock CpGs" from our dataset and computed the age of our study participants using the tool's identified regression model.
To assess the impact of marker density on C-kernel performance, we sampled 10 times a subset of markers on the maize panel dataset and computed the log-likelihood ratio test for the C-kernel.
Using a simple Rescorla Wagner learning rule, with two learning rates (for CS+ and CS−), we fitted each participant's dataset and computed the explained variance under an optimal learning rate (see Section 2 for details).
Since no data from the normal tissue were available for this dataset, we extracted normal tissue data from dataset GSE89, and computed the mean log2 transformed ratios of BC vs. these extracted data, which we used as controls (Table S2).
The results for ranking the markers based on the USA dataset and computing the scores using the Netherlands set are shown in Fig. 4D.
For a motif, given the number of matches in the foreground dataset (ORGs), the p-value corresponding to the motif occurrence was estimated based on 1000 random samplings of the background dataset and computing the fraction of times that the motif occurence equaled or exceeded its occurrences in the foreground dataset.
