4

To start, let me say i am a programmer and not a math wiz, so this question might be very simple.

I have a data set of prices that sort of looks like this

1399.00
13.99
13.97
13.83
13.48
...
~9

So what happens is everyone "bids" a few cents lower than everyone else, and one person had an "error" that resulted in 1399 isk instead of 13.99 (I cannot sanitize the data, read on)

The data is presented to me in the following XML format:

<marketstat><type id="34">
<buy>
<volume>16831614476</volume>
<avg>4.60</avg>
<max>4.73</max>
<min>2.96</min>
<stddev>0.45</stddev>
<median>4.69</median>
<percentile>4.72</percentile>
</buy>

Using only these data points, is it possible to predict the "true" max, ie discard the ridiculously high price and provide a value near the "max"?

Mike Buck
  • 41
  • Sorry, so to be clear: you're given the information from the buy-statement, but not the data itself. Is that correct? If you do have access to the data, why can't you "sanitize" it? – Ben Grossmann Oct 28 '13 at 22:44
  • I don't see your problem. In your XML example, a median above the average seems to indicate rather that there are errors at the lower end instead of the upper end. And of course the fact that the max is not at least ten times the min indicates that probably no typoes have occured. – Hagen von Eitzen Oct 28 '13 at 22:50
  • This is quite similar to my question here: http://stackoverflow.com/questions/19269638 – Krista K Oct 29 '13 at 05:26

0 Answers0