# Difference of unequal sample size

I am a newbie in statistics.

I have two columns of numerical data of unequal length, motion data. I need difference of the two columns, by shortening the longer column.

766.5 543.8
764.9 544.3
764.0 544.0
765.5 543.7
764.0 543.2
764.4
765.3
765.8


Now I want a third column of data which is the subtraction of column two from column one. So I want to re-size column one data to the length of column two, either by deleting some values or something like that. i.e there are 8 values in first column and 5 values in second column. So I want to resize column one to length 5, with minimal loss in the SD and other parameters. I want to use the results of the difference of two columns as a input to neural network and try to build a predictive model with this data.

Are there any methods / algorithms which can help me achieve this, while keeping the accuracy of the shortened column good enough?

• Perhaps if you stated why you're subtracting them and what makes subtracting the first two values of the columns any different from subtracting the first value in column A from the 6th value in column B. – John May 10 '14 at 16:44
• Since these are motion data, the time information is essential for their analysis. Do you have explicit times for each observation or are the times at least implicit in the sequence of values (based on some understanding of a common time interval between each pair of successive values)? – whuber May 10 '14 at 17:55