Media stream quality is highly dependent on underlying network conditions, but identifying scalable, unambiguous metrics to discern the user-perceived quality of a media stream in the face of network congestion is a challenging problem. User-perceived quality can be approximated through the use of carefully chosen application layer metrics, precluding the need to poll users directly. We discuss the use of data mining prediction techniques to analyze application layer metrics to determine user-perceived quality ratings on media streams. We show that several such prediction techniques are able to assign correct (within a small tolerance) quality ratings to streams with a high degree of accuracy. The time it takes to train and tune the predictors and perform the actual prediction are short enough to make such a strategy feasible to be executed in real time and on real computer networks.

Supervised learning is a classic data mining problem where one wishes to be be able to predict an output value associated with a particular input vector. We present a new twist on this classic problem where, instead of having the training set contain an individual output value for each input vector, the output values in the training set are only given in aggregate over a number of input vectors. This new problem arose from a particular need in learning on mass spectrometry data, but could easily apply to situations when data has been aggregated in order to maintain privacy. We provide a formal description of this new problem for both classification and regression. We then examine how k-nearest neighbor, neural networks, and support vector machines can be adapted for this problem.