2

As I'm not a mathematician I thought I'd ask here for advice how to approach something I'm working on (probably a basic question for a lot of you guys, but it was a subject of a debate at my work earlier)

We have this graph: price trend graph and here we're trying to show the prices trend for products in "Shoes" category, for the last year. Every week we extracted the average discount for the ending week, for all the products available at that time in the "shoes" category, and now we put them in this graph. For example during the week ending on 30 June 2013 we had an average discount of 9%.

This graph shows us very easily the price trends, the discounts periods, the discounts magnitude, for the last year, but it is also misleading because comparing the starting point with the ending point it states that overall we had a discount of almost 50%, which is not true all the time (because new products arrived with bigger prices, and the discounted ones have been sold). Is there any better graph representation, so we keep the price trending but not to mislead with the values?

What We've tried

In order to make it more understandable by the users I've generated a second graph to show the discounts level evolution (summed up for 4 continuous weeks, you can see it at http://vetements-et-chaussures.franceprix.fr/#evolution_prix, the second graph) but the discounts level graph doesn't actually really picture the big picture of the prices general trend (like in our first graph).

Any suggestions how a problem like this could be approached? All ideas are more than welcome.

Stefan29
  • 21
  • It seems that a linear correlation could be appropriate as first approximation. – Claude Leibovici Mar 04 '14 at 09:20
  • If you are able to post a few numbers taken from the graph (percent vs date), I should try. I cannot do it by myself since I am almost blind. – Claude Leibovici Mar 04 '14 at 10:25
  • Hi @ClaudeLeibovici ! Thank you for your interest. Being all about a visual graphic to express statistical data, I am afraid that this might be problematic to you. The real data is not so important, we can have also an easier scenario: Let's suppose the products in the "shoes" category are discounting each week with 5%. After 12 weeks the total discount gets around 54%. In this case my graphic will show the price being constantly discounted every week with 5% (and that's good!). But I don't want the graphic to suggest an overall discount of 54% for all the products, as that may not be true. – Stefan29 Mar 04 '14 at 12:05

0 Answers0