4 ms·
The values that go into the formula are normalized, but not weighted. So a similar screen size is "worth" as much as a similar weight. If you find examples, whe
by no_gravity 11y ago
The values that go into the formula are normalized, but not weighted. So a similar screen size is "worth" as much as a similar weight. If you find examples, where weighted results would perform better, let me know. You can compare the given phone to other phones by clicking them in the list on the right.
- leni536 11y agoNormalized to what? For me normalization or weighting is equivalent. Putting it this way there could be different normalization factors. Examples: 1. Average value over all phones 2. Standard deviation of the value over all phones 3. Minimal noticeable difference in the given property (it's quite vague though, it's basically custom weighting factors) I don't think that there is an obvious best, but I don't think it's needed and all gives sensible results. A more sophisticated distance wouldn't be Euclidean. If you look at x-resolutions and y-resolutions independently you won't catch exotic aspect ratios which should be weighted more compared to a standard aspect ratio phone.