First of all I have to mention, that I'm not a statician at all, I'm just a simple programmer and I have some curiosities... and the wrost of all, i don't know where to start from.
Let's assume the following working scenario:
A big company, a internet service provider (ISP) with unlimitd bandwith choose to change how the users will use and pay for the internet services: each user has to predict how much bandwidth they will consume in the next day for each hour. If the user predict that she will consume 0MB from 0.00 TO 18.00 she will pay nothing. If the user predict that she will see a HD movie from 18.00 to 20.00 they will comsume 10GB and they will pay just for that ammout of data.
If the user consume more that she predicted, she will pay more for that ammount of data (just for the differece). Predicted mount of data is cheapest. If they consume more that it was predicted, they will have to pay penalities.
The thing is that the users can build networks/groups with they freinds in order to optimize their costs. For examaple, if a user is not using his predicted amount of data, another friend from theyr network can use it, for free. If they wish the users of a group, can see (at every 30minutes) if their freinds are consuming traffic or not.
Idea is that each user has to predict how much internet they will comsume, based of theyr habits and schedules. The users will pay at the end of each month the trafice that they consumed.
Now, the quest is to find the most appropiate way to represent the traffic prognoses for each user, group and for the ISP.
- I'll use line charts to represent how much user has predicted and how much she consumed in each day for each hour.
- I'll use bar charts to represent the difference between what was predicted and what was consumed in each hour.
The problem is to represent those data in a graphical way for each month, for each user, group and the ISP:
-Can you give me some samples or resources which can help to find the best way of represeting my data costs?
-Do you have an idea of what statistical approach to use in order to illustrate the prognoses, comsumes and errors?