Statistics with Team Kodi values - Printable Version +- Kodi Community Forum (https://forum.kodi.tv) +-- Forum: Discussions (https://forum.kodi.tv/forumdisplay.php?fid=222) +--- Forum: Kodi related discussions (https://forum.kodi.tv/forumdisplay.php?fid=6) +--- Thread: Statistics with Team Kodi values (/showthread.php?tid=255120) |
Statistics with Team Kodi values - Tolriq - 2016-01-06 Since a friend send me a interesting analysis I thought I have to post it Razee stats : Kodi 15.2 5416841 Kodi 14.2 1690940 Kodi 15.1 1108620 XBMC 13.2 465756 Kodi 14.1 378745 Kodi 15.0 301048 Kodi 14.2-RC1 216801 Kodi 14.0 204483 XBMC 13.1 104436 Kodi 15.2-RC1 89776 This means a population of 9977446 users for this study. If we take those numbers : 31% of users have not updated post 14.2 Now let's do plain math as with Koying facts : There's 3 million Kodi Android users for sure and 98% of them are 15.1+ So let's analyse data of non Android devices. This leaves to 44% of non Android devices are not updated past 14.2 Now let's take another Koying less sure fact : 50% of Kodi users are Android This leaves to 60% of non Android devices are not updated past 14.2 All those numbers not related to Yatse or Tolriq just values that you have. I know every message I try to pass are either unwanted or not understood but please forget about Tolriq, about Yatse, about everything You have some numbers and surely tons of others hidden, please take time to analyse them as you may find tons of interesting things. RE: Statistics with Team Kodi values - wsnipex - 2016-01-06 here are some more OS details on the stats we have based on unique downloads of the current version of the tmdb scraper: in Month January 2016 Code: os Downloads platforms: (Build xxxx is probably some windows stuff, need to investigate) Code: platform Downloads Versions Code: browser Downloads RE: Statistics with Team Kodi values - Tolriq - 2016-01-06 So 65% of Android, meaning that if the Razee numbers are true this is way worse about other platforms. But I do not think that there was an update of TMDB in January so this more represent the re-partition of OS for new install than current user base. But this is still a correct indicator of the future repartition. I was just trying to pass a message that I know is true from user feedback, but with numbers to try to prove factual. I did not expect such rejection of the possibility of the idea, but I really hope you do take time to analyse your own data and may be find the message yourselves RE: Statistics with Team Kodi values - wsnipex - 2016-01-06 you always have to take those numbers with a grain of salt, i.e. which boxes are used most, are always on etc. Early in the month, always on boxes obviously will update asap, while other installs take a bit to catch up. Those stats point a general direction, but based on past months its ~50-55% android. And I do get your point. Sadly we don't have a direct mapping of kodi version to platform. I'll put that on my todo list. RE: Statistics with Team Kodi values - Tolriq - 2016-01-06 Thanks About update yes, that's why I do use 30 days sliding window for my own stats. I consider active user a user that use Yatse at least once a month. Stats will be interesting for next update, or accumulated for December / January since last update seems to be half December. RE: Statistics with Team Kodi values - wsnipex - 2016-01-06 this is December 2015, but you've already seen those from razzee Code: browser Downloads RE: Statistics with Team Kodi values - Tolriq - 2016-01-06 Not the platform re-partition ones to use the correct Android ratio to be able to deduce the non update rate of non Android devices. But if we keep the 50% suggested rate from previous month then the seeked value is : 60% of non Android devices are not updated past 14.2 RE: Statistics with Team Kodi values - trogggy - 2016-01-06 (2016-01-06, 12:00)wsnipex Wrote: you always have to take those numbers with a grain of salt, i.e. which boxes are used most, are always on etc.What I find really surprising about those numbers is how low they are for openelec - especially as many of them will be always on (raspberries). RE: Statistics with Team Kodi values - phate89 - 2016-01-06 Imho all this numbers should be taken very very carefully. Statistically speaking both methods have no way to state how accurate they are to represent the kodi userbase universe set so there's no actual way to see how accurate they are. Plus you don't know the reasons behind this numbers. No new features in 15? No bugs to fix in helix? More addons lost in 15? Without automatic update we don't have a good method to know if they're just lazy or what.. RE: Statistics with Team Kodi values - Tolriq - 2016-01-06 (2016-01-06, 13:01)phate89 Wrote: Imho all this numbers should be taken very very carefully. Statistically speaking both methods have no way to state how accurate they are to represent the kodi userbase universe set so there's no actual way to see how accurate they are. You mix 2 things : First yes no data are perfectly accurate but when multiple different methods shows the same result then there's no smoke without fire. Second : This is all the .... question : Why do users do not update ? I have feedback from users but no one believe, so I just try to show that this is the question to look at. RE: Statistics with Team Kodi values - wsnipex - 2016-01-06 What from I gather from the forums, which is arguably not alway representative of the whole user base, the primary 3 reasons to not upgrade are: in no particular order: - current version works fine as is <- thats fine, look at the millions of Win XP for example - afraid of bug in new releases, specially .0 <- hen and egg problem. We need users testing, but not too many report bugs - piracy addon breakage <- nothing we can do or care about RE: Statistics with Team Kodi values - Tolriq - 2016-01-06 From my specific user case feedback there's also an increasing cases of - New version works less good than previous for example : http://forum.kodi.tv/showthread.php?tid=254686&pid=2204557#pid2204557 About the bugs, from the feedback I have the 14 Windows crash bug with remotes was a deal breaker for a lot of people and the way it was handled have made that users are now more regarding about updates. I do not say it was an easy problem to solve or that release should have been postponed for a known Windows breaker problem, but there was at least a communication problem around that. And of course this specific problem about remotes is more related to my user than the generic population but there's a lot of users that use JSON remotes not only Yatse ones. But generally confidence in new version have decreased at least on my specific population. About bug reports I did proposed a long time ago and some others too if I remember the creation of a special group for advanced / recognized users and / or externals devs so they can make proper bug reports without the noise and that could be easier to find for the Team and to analyse with easier communication. Because other ways post are lost in the forums and Trac was never really correctly used and is spammed. RE: Statistics with Team Kodi values - trogggy - 2016-01-06 (2016-01-06, 13:20)wsnipex Wrote: - afraid of bug in new releases, specially .0 <- hen and egg problem. We need users testing, but not too many report bugsYou don't make it easy tbh. I'm reluctant to 'upgrade' blindly - but I'm keen to try new releases. Once I'm confident everything's working I move to the new release. On windows it's easy to run a portable install to test a new release. I doubt many users are aware of that - I would think (but don't know!) it would be fairly straightforward to give the option of a portable build when installing, or even to give an option to do a portable build and copy (not cut) existing data across. On Android it's a complete PITA to run a nightly (or new release) alongside an existing build. Again, that doesn't have to be the case. In openelec it's quite possible to test a new release alongside an existing build - but it's not documented. They're the three platforms I use, can't comment on others. But in any case there isn't a straightforward, well-advertised way to comment on bugs here. RE: Statistics with Team Kodi values - Prof Yaffle - 2016-01-06 I'm happy to actively test lots of things, and report bugs where I see them. However, I'm far more conservative with Kodi than with anything else (well, except my server OS). I'm wondering why... 1. Because I use OpenELEC as the main clients, and my family relies on those to work at the touch of a button. 2. Because I user a shared mysql database, and there's no easy way to keep multiple versions in sync. 3. Because I also have Windows (not used much) and Android (used a lot) clients that need to fit in. So, while I built Jarvis on my Linux system and ran it standalone 'for a look', there's no easy way to put it into 'production' on this box without breaking everything for my family. And that would be ugly. I suspect there are three things, then, that would make it easier for me as a user rather than as a would-be techie: 1. Some kind of sync system between MyVideosNN and MyVideosNN-1 (et al) so you don't need a homogeneous version base. And no, UPnP library sharing isn't a replacement, it's horrible. 2. A better way to run a new/test version in parallel (portable mode) without breaking the main system - including addons (I didn't enjoy cmaking my own PVR addons...) 3. Perhaps an in-app bug reporting system of some kind - any kind? Maybe linked to the log uploader addon, so you could guarantee a basic set of information alongside a bug report - perhaps even automated if there's been a crash (e.g. a crashlog is found)? I have logged stuff on trac, but it's not the most friendly way to find/report things, and the confusion between feature requests here and bugs there vs github bugs for addons is probably too much for many. ---- As for the statistics... the problem I have is that none of these is a typical, representative sample. They're all random, or self-selected, and I think that's where the differences of opinion come from. Yes, Tolriq is right to say that enough randomness will ultimately converge to reality (the monte carlo effect in any statistical model), but that does mean looking at all significant data sources and trying to overlay them, as any one of them isn't going to be right: I never use Yatse to control my desktop or Android tablet, for example, and rarely even to control our main TV as I use CEC - but I use it all the time to see/control what the kids are doing in the next room without opening the door... Anyway, not trying to insert myself into someone else's conversation, as we've had too much of that on other threads recently. I thought I'd just add a comment from the perspective of a dumb user. RE: Statistics with Team Kodi values - RockerC - 2016-01-06 (2016-01-06, 12:06)Tolriq Wrote: I consider active user a user that use Yatse at least once a month.How do you track which are active users and which are passive users? I have Yatse as well as many other remote control apps for Kodi installed on both my phone and tablet which are both set to auto-update apps on new releases, but I'm definitely not an active Yatse user, or an active user of any other remote control apps either for that matter. I only use a physical remote on a daily basis, and maybe start Yatse and other remote apps once or twice a year at a party just to show people what more they can have with Kodi. Currently my main device in the living-room for Kodi is a Raspberry Pi and I use the television sets remote via http://kodi.wiki/view/CEC but I have have HTPCs and other devices with Kodi that use a dedicated IR or RF remote control. While I have no stats I don't believe I am an exception today to having loads of apps installed on my phone and tablet that I hardly ever use, if ever, once installed and tested for the first time. I even have loads of apps and games installed that I have never even started or tried once but have only installed because I heard good things about them and eventually planned to try them out once have have the time, time which I almost never make and take for trying out new apps and games. |