We investigate the application of machine learning methods in the domain of tunnel construction. In tunneling the use of tunnel boring machines (TBMs) is often preferred to the drill-and-blast (d&b) method, mainly due to higher advance rates. However, a particular disadvantage of shielded TBMs is that the encountered rock at the tunnel face cannot be examined visually, but sensors of the TBM continuously record data about the machine status during driving. For the first time we investigate if the sampled sensor data have sufficient discriminative power to classify geological parameters automatically by statistical pattern recognition using Parzen windows. Experimental results show that certain types of geological formations can be recognized with high accuracy.