Furthermore, we also performed the expression analysis based on the mixed datasets.
Despite the negative correlations, this location was included in the mixed datasets produced from the site-specific model selection.
The performance deteriorates for the mixed dataset, although peaks were formed, as shown in Figure 4.
The mixed dataset contained 6,486 tweets, with 6,091 containing no cyber hate and 395 containing cyber hate.
We found that (i) the use of the mixed dataset with a Mann-Whitney p-value cutoff for feature selection could discard source-specific features, and (ii) a quantile normalization step helps to select (for machine learning) partially concordant features, in which the trends are concordant between sources, but intensity levels are different between sources.
The use of the mixed dataset with a Mann-Whitney p-value cutoff for feature selection could discard source-specific phenomena.
In the second scenario, the PIR1 and PIR2 datasets are first mixed together and then the training and testing datasets with 949 and 534 samples, respectively, are randomly generated from the mixed dataset.
Further expression analysis was performed based on the 4 paired datasets and mixed datasets, respectively.
Next, we compiled mixed datasets for testing the read-mapping pipeline, by randomly sampling subsets of these simulated sequence reads.
Simultaneously, due to dynamic expression across different individuals, equally mixed datasets were also used to estimate the expression patterns.
