Generalization of point data

I have a large set of spatially distributed point data with certain values. I need to reduce the number of points. The basis of reduction is to identify points that are within a proximity of, say 100 meters. If more than one point exists, then a new point (located at its center / centroid) should be created which should now have the mean of all points that have been merged.

Can I do this without serious programming? May be a long but workable solution using available tool sets?