Round off error in the database?
Originally posted by Jason@Sep 9th 2004 @ 7:58 PM
Ah. Looks like that was the problem. I reset the variables when the script looped (recalculating all of the overall mileages) and it appears to have worked. Please double-check me on that.
Ah. Looks like that was the problem. I reset the variables when the script looped (recalculating all of the overall mileages) and it appears to have worked. Please double-check me on that.
Andy, thanks for being such a careful observer!
Andy
P.S. Oh, I was really busy at work yesterday. I was going to take a look at your code fragment last night but my home PC blow up. Took me all night to find out it's the video card and swap in an old card.
Hi Jason:
___I thought we had discussed the lmpg inaccuracy a few months ago with Rick Reese’s and my own tanks but I didn’t find the thread? In any case, my current spreadsheet lmpg now matches up with the database lmpg almost exactly. Wait until I post my next tank at which time my spreadsheet, my lmpg, and the database will finally match because it is going to be a good one
___Accwai, thanks for leaning on Jason for the accuracy fix. I always thought it was a significant digit problem and kept hammering on Jason to fix that to make the lmpg numbers match. Significant Database Updates, By Popular Request * tenth post + down …
___Good Luck
___Wayne R. Gerdes
___Hunt Club Farms Landscaping Ltd.
___Waynegerdes@earthlink.net
___I thought we had discussed the lmpg inaccuracy a few months ago with Rick Reese’s and my own tanks but I didn’t find the thread? In any case, my current spreadsheet lmpg now matches up with the database lmpg almost exactly. Wait until I post my next tank at which time my spreadsheet, my lmpg, and the database will finally match because it is going to be a good one

___Accwai, thanks for leaning on Jason for the accuracy fix. I always thought it was a significant digit problem and kept hammering on Jason to fix that to make the lmpg numbers match. Significant Database Updates, By Popular Request * tenth post + down …
___Good Luck
___Wayne R. Gerdes
___Hunt Club Farms Landscaping Ltd.
___Waynegerdes@earthlink.net
Right, I remember that. I think the error was blamed on rounding, however. Unless I don't recall correctly, I assumed that whole-number distances and tenths-digit mileages was creating the abberance.
Ah, well. No reason to dwell on the past. It's fixed.
Ah, well. No reason to dwell on the past. It's fixed.
Originally posted by Jason@Sep 10th 2004 @ 11:56 PM
Looks like things aren't absolutely perfect after all... I see a "highway" percentage in the 90's while every tank's percentage was 0-33. I'll look into it. These are sort of useless variables, anyway
Looks like things aren't absolutely perfect after all... I see a "highway" percentage in the 90's while every tank's percentage was 0-33. I'll look into it. These are sort of useless variables, anyway
Andy, for the life of me I cannot figure out why these values are wrong. If it works for mileage, why not the others, as well?
Code:
if ($weight['mileage'] != 0) { $mileagedenominator = $mileagedenominator + $weight['distance']/$weight['mileage']; }
if ($weight['temperature'] != 0) { $temperaturedenominator = $temperaturedenominator + $weight['distance']/$weight['temperature']; }
if ($weight['time'] != 0) { $timedenominator = $timedenominator + $weight['distance']/$weight['time']; }
if ($weight['highway'] != 0) { $highwaydenominator = $highwaydenominator + $weight['distance']/$weight['highway']; }
if ($weight['traffic'] != 0) { $trafficdenominator = $trafficdenominator + $weight['distance']/$weight['traffic']; }
$numerator= $numerator + $weight['distance'];
if ($mileagedenominator != 0) { $mileage = $numerator / $mileagedenominator; }
if ($temperaturedenominator != 0) { $temperature = round($numerator / $temperaturedenominator,0); }
if ($timedenominator != 0) { $time = round($numerator / $timedenominator,0); }
if ($highwaydenominator != 0) { $highway = $numerator / $highwaydenominator; }
if ($trafficdenominator != 0) { $traffic = $numerator / $trafficdenominator; }
Side note: the main bar graph was running on numbers rounded to the tenth insteaded to the whole number. The drawing software rounded these values down. This has been fixed, resulting in a Prius II average of 50 instead of 49.
Side note #2: I am contemplating changing these values to medians instead of means in order to dissassociate "hypermilers."
Side note #2: I am contemplating changing these values to medians instead of means in order to dissassociate "hypermilers."
Thread
Topic Starter
Forum
Replies
Last Post





