These results validated the generalizability of the GPR modeling approach, as high accuracy was achieved when training the model on one data set and testing on an entirely independent data set.
Both databases can be used in combination to verify hypotheses, to cross-check a finding with one database on the other, or to train a mathematical model on one data set and to test it on the other, thereby avoiding the issue of overfitting.
Here, we train the parameters of our model on one data set and evaluate its performance on the other data set.
Bias becomes evident when testing an algorithm's cross-data set generalization ability, or, in other words, training a model on one data set and applying it to another.
We trained each model on data for one TF and used it to predict the binding specificity of related TFs.
An alternative approach that is somewhat more stringent is to develop a statistical model with one data set and test the replicability of precisely that model with other data.
> -wrap-foot> Westimatethethe influence of each network on our integration model by comparing the quality of the initial model (consisting of four networks and their corresponding GDV similarity matrices) with the quality of the model with one data source removed from the initial set.
Fitting the multivariate model to one data set (Oslo2) and subsequently predicting protein expression in the two other data sets, more than half of the proteins that were present in all data sets resulted in significant predictions.
As a result, the prediction models built on one data set show a significant deterioration of their accuracy when they are used to classify new, unseen data.
We will use deviance differences to quantify the additional lack-of-fit when a model is fitted on one data set and predictions are made on another data set [ 53].
