There are a lot of things you can do and it depends on your expertise with statistics. You could start with moran I (simple in arcgis, GEODA is a free software that is a better way of calculating moran) to see if there is an overall trend of autocorrelation in house prices, ie high priced houses are near one another or not, or house prices are random.
What I think is better and getting closer to what you asked would be a regression type question. To do this I would calculate the distance of each house to a train station and test for a relationship. House Price ~ Distance to train. If there is a slope in the relationship you are golden. However, in doing this analysis you will need to consider the fact that the driving factor behind the price of a house is not the distance it lies from the train but other things such as age, size, condition, yard, view etc. These factors will act as confounds to your analysis, and I expect any relationship to be somewhat weak if you can not control for these factors. Again I would suggest you do regression type analysis is a stats package.
David
