standardization with positive range

[Devashish Singh]

i want to standardize a data set such that all of the resulting values are strictly greater than 0. i cannot use min/max method as it will include a 0 value. usual standardization method(subtracting mean and dividing by S.D) is giving me some negative values. can you please suggest some other method?

[Nick Cox]

Any number of ways of doing this such as value MINUS minimum PLUS smidgen. But some insight into why you want to do this might imply a better answer. There needs to be a way to explain smidgen as more than ad hoc.

[Devashish Singh]

i am creating an index. i am using geometric aggregation to aggregate different indicators. however the indicator values need to be normalized before doing that. but geometric aggregation method can only be used if the normalized indicator values are greater than zero

[Nick Cox]

That explains more of why you want this, but unfortunately I don't have a better answer. If what you want to do involves taking logarithms or equivalently ratios, then you are going to get different results with log (x + c) and log (x + d) if c and d are different constants each ensuring x + c > 0 and x + d > 0. This can't be a new problem and should be discussed in the literature you are using.

[Devashish Singh]

thank you. i will refer to the literature