Message Boards Message Boards

0
|
10512 Views
|
1 Reply
|
0 Total Likes
View groups...
Share
Share this post:

Modeling bimodal spending patterns

Posted 12 years ago
I have been looking at spending patterns. The data systems I have been looking at are nearly always bi-modal. There tends to be a lot of people spending on the low and then a group of people that spend on the high end. This is an example:




Given similar spending contexts the data is visually quite consistent.

 I am looking for a simple way to fit an equation to these curves and get fit error indicators.

My assumption is that given the person's income (which I do not know) they are likely to contribute to a lower normal distribution or to an upper normal distribution.

I would like to know how I might confirm that assumption. And how I might find other equations that fit the sets better than two normal distributions. I have found ways to come close with a lot of trial and error, but it seems like there should be an easy, slick way to pull apart problems like these. Any suggestions on a general approach would be greatly appreciated. I would like an equation that would allow me to simulate the spending behaviors in the different contexts.
POSTED BY: dale Strickler
How about assuming a normal distribution whose mean depends on the individual's income and then using deconvolution to to get the income distribution?
POSTED BY: Frank Kampas
Reply to this discussion
Community posts can be styled and formatted using the Markdown syntax.
Reply Preview
Attachments
Remove
or Discard

Group Abstract Group Abstract