Evaluation of impact over time


New Member
I have 3 groups and their data: incoming customers incoming funds

When looking on 2 week resolution, I check the percentage difference and can evaluate the impact on these parameters.

If I have 5 weeks, data for each week, I need a better statistic to evaluate the impact on the parameters in question (Customers, Funds), otherwise I get just 4 values that represent the change from week to week.


Fortran must die
I am not sure what question you are trying to answer. Are there changes over time, do certain effects influence some variable over time etc. Why can't you simply measure the starting value at week one and the last value at week 5 and ignore the intervening values?


New Member
I will try to explain myself better.
I want to find a better measurement for changes,rather than percentage change. my purpose is to discover anomalies in the data from week to week. (or even in "stronger" populations).
would you recommend using perhaps some other statistic value? Q3-Q1? 95 percentile? mode?