Modern cellular networks utilising the long-term evolution (LTE) and the coming 5G set of standards face an ever-increasing demand for low-latency mobile data from connected devices. Header compression is employed to minimise the overhead for IP-based cellular network traffic, thereby decreasing the overall bandwidth usage and, subsequently, transmission delays. Since Robust Header Compression, among others, is primarily designed for the compression of live audio transmissions on endpoint devices, it performs best if certain fields, like the IP ID and RTP Timestamp, stay constant or change at a predetermined rate. Moreover, the compressor expects the uncompressed stream to be loss-free as well, which might not be the case in general scenarios. We employ machine learning approaches for the prediction of Robust Header Compression version 1's and version 2's compression utility under various loss rates and header field dynamics. We analyse how the compressions react to different fluctuations in the headers and choose the compressor configuration which maximises utility. We show that the appropriate choice of compressor repetition configuration increases the overall utility under dynamic channel conditions, such as in the case of remote steering and various IoT applications, and finding the optimal configuration could produce significant benefits, like 1.2 speed-up in a fully utilised network.