# What should I do to compare different sets of data?

I am a beginner in statistics, and I want to learn machine learning :). Therefore, I have gathered some sample data to practice. But, the problem is I want to create a feature (or attribute), which is common for every entity. But, this feature is observed and measured for entity X in Y meters and Z seconds, and for entity W in M meters and L seconds, and there are many observations like described before. Which algorithm should I apply to create a common feature for all entities? BTW, I am using WEKA.

-
Could you clarify what you mean by a "common entity". It's not clear to me because it seems you are describing two features (for me, a feature or attribute = a variable) with different units (meters and seconds) rather than a single one? –  chl Jan 31 '11 at 18:25
@chl with common entity, I mean an "instance". As I tried to explain above, there are several measurements with different criteria(s), I need to normalize(?) them to further use in regression algorithms. But, I am not sure how to make these numerical measurements to carry same meaning for each instance. –  baris_a Jan 31 '11 at 19:27
Could you give an example of sth that is recorded as a single entity in meters and seconds? –  chl Jan 31 '11 at 21:48
@chl assume there is a runner, who runs 100 m in 15 secs, and 200 m in 32 secs, other runner runs 500 m 40 secs and 1000 m 83 secs, who will run 1 m (or a common) distance better? is this process related with standardization or normalization? It may seem simple but we can also assume that running 1000 m is harder than 100 m. –  baris_a Feb 1 '11 at 11:39
Can you give us example data? arff file or csv file? –  Atilla Ozgur Apr 12 '12 at 8:13