Welcome to Netrider ... Connecting Riders!

Interested in talking motorbikes with a terrific community of riders?
Signup (it's quick and free) to join the discussions and access the full suite of tools and information that Netrider has to offer.

Building a 'Which Bike For Me?' Quiz

Discussion in 'General Motorcycling Discussion' at netrider.net.au started by Bravus, Apr 1, 2008.

  1. Hey all

    In reading all the 'which bike for me?' posts in the Reviews and Suggestions forum I was thinking that it would actually be possible to build a survey that asks someone say 20 questions about their needs and interests and experience and finances and so on and then spits out one, or perhaps a list of 3 or so, recommendations of possible bikes that fit their needs.

    It would be easy to do a humorous one ('how much do you spend each week on lattes?' ;)) but not that much harder to do a fairly serious one that was actually useful. Of course it would be biased and all that, but still...

    My question is just what platform or set of tools would work best. Something like the incessant quizzes on Facebook might have the power and flexibility to do it simply, but any other suggestions would be very welcome.

    It's something I could see myself putting the work in on if the tech would support it and people thought it could be useful.
  2. Good idea.

    Should be quite easy to get this started, all you would need is an anonymous survey for existing owners asking the same 20 questions but with also a question on which make and model of bike they have. This will build you a database of profiles against each bike.

    Then when a potential buyer answers the same 20 questions the system could do a best match to existing profiles. You could also profile other bikes so they could appear as other suggestions.

    Best if each question is multi choice so that matching is easier.

  3. Thats bizzare, I was thinking of building exactly this system for my flash project this semester! :S

    I'm not entirely sure that I'll be able to do it as I dont know actionscript very well yet.. I could do it in C or vbasic or something.. Its something I'll be thinking about in the next few weeks.

    edit: My idea for how it would work would start with a giant flowchart of sorts and work out some logic systems using that.
  4. i would be too afraid to take the quiz incase the results would suggest that i buy a vespa! :shock:
  5. What's wrong with a Vespa?
  6. I think the statistical approach that Rybky suggests could work, but that (a) it'd need a fairly huge sample size to be really useful and (b) a lot of people have chosen a bike that is not really ideal for their needs based on factors like budget, other members of the household, taste (or lack of same!) and so on...

    I was imagining more like a heirarchical sort like phizog was thinking of. I thought about a few questions as I was riding in this morning.

    If you started with 'What state are you in?' and 'What kind of license do you have?', that would immediately cut it down to 250s in some states and LAMs bikes in others for some people, which would be a decent place to start sorting.

    Then you'd need to ask about budget, most common uses, preference for faired or naked, cruiser or sport, how often (if ever) a pillion is carried. Maybe even questions about rider height and weight.

    Phizog, I'm totally happy to cede the idea to you entirely and sit back throwing (constructive) stones from afar!

    Other suggestions for questions for questions would be very welcome, but realising that it will have to be whittled down to a managable size survey (20 was just a number out of the air but 60 is silly, for example).

    So for me, in QLD, full license, prefer something sporty, half-faired, pretty beefy and pillioning a lot, it could spit out a list like the Bandit, ZX9R, ST and GTR and so on. You could even add the extra sophistication of a ranking based on different criteria: ST if you want to look like a cop, Bandit if you pillion a lot and feel that your arms need lengthening. ;)
  7. http://www.okcupid.com/tests/take?testid=17525870837436152068

    unfortunately it's been taken down. It was a good quiz ...

    EDIT: this question now irrelevant -- Are you aiming to cater to LAMS / non-LAMS learner / unrestricted riders ?

    Have you thought about tying it into that MC fuel economy project someone was talking about recently? Economy is obviously a factor in most people's purchases, anyway.

    Phizog: I reckon flash is the wrong tool for the job personally. If it *can* be done with vanilla HTML forms, it's almost always *best* done with HTML forms. Think of something it's impossible to do with plain HTML ftw ... like a drag & drop customization thingy, or something ... my 2c only
  8. I'd prolly do it with a weighted points system:

    each answer-MC relationship (many-many .. ) has a points value associated with it, which can be positive or negative (or null / zero, if the question has no relevance to the machine in question).

    The three or so highest scoring bikes at the end are shown.
  9. It is entirely possible to make a good one using data mining, but NetRider admins wouldn't be too happy about hosting an application which takes up most of the resources of their server once every few minutes (i.e. when someone actually uses it. That's one issue. Another one is that it'll be impossible to tune the mining algorythm properly until there are at least several hundred responses, which will render the system completely useless at the early stages.

    Having said that, I think it's a great idea and I, for one, am very much interested in turning it into an actual system. I have the means and time to do it. The question is how many people would be interested in actually using it :)

    jekyll, what you're proposing sounds a bit like a Neural Network. That one will f@#k up pretty much any server there is :)
  10. Well, I did also speculate about the possibility of building it as a commercial web-based system, free to the user but funded by ads and possibly by dealers who pay for referrals... that's a whole other level from a simple free and open-source tool, which would be my first preference.
  11. Some questions I'd been thinking of include:
    - Do you need to regularly sit on freeway speeds?
    - Height + Weight
    - How important fuel consumption is to you
    - Budget
    - New or used
    - Need to pillion on this bike at some point?
    - What riding position do you want, upright and relaxed or leaning over (sporty)
    - Interested in riding fast or just cruising
    - Etc.

    I think a hierarchy system would be the easiest to work out, basically a big stream of if statements..
  12. Hey, I'm a student, I don't do commercial :grin: Plus it's a whole other level of commitment, which will take the fun out of it. Someone else could be interested though, but I honestly can't see it as a viable project at this stage. If you have a database of 10,000 responses, it is possible to make good money on it, but that's pretty far off at this point. I was rather thinking free, simple and dirty :)

    phizog, I take it you're a student too? :) If you're good with flash, you're probably good with design/graphics in general. We could join forces if you're interested. I'm a final semester BScIT (Software Engineering) @ UTS
  13. Sweet, yes I'm a student doing digital media. I started with computer science but the maths got to me! Teaming up sounds good.
  14. This is not a new idea. A mate and I have been working for a little while on something with the ability to do this. The two biggest issues deciding what information is important in making these decisions and then how to go about collecting it. Once you work those things out, it's not a tough project.

    I'm not sure how putting some commitment into something takes all the fun out of it.

    In my opinion that certainly would take the fun out of it. If a job is worth being done, it's worth being done properly.

    No, it sounds like a database which any server should handle efficiently, even if you had 10,000 bikes in your database.

    Right on, I see no need for anything fancy on the frontend at all.
  15. Yes I know flash isn't the best tool for it, but I've got to do a project using it :) I'll have to talk to my tutor about how complex it would be to do.. if I can use the logic I'm thinking of, it shouldn't be too intense at all, and its just a project so it wouldn't have to be full scale, just something with enough sample data to get the general idea. Eg not EVERY learner bike, maybe just about 50.
  16. If this goes ahead please make sure to include photos of the bike models as well. In NSW we have the LAMS list to choose from but on the RTA site it is just a list of manufacturers and models, no pic or description of the bike! So for a newbie to bikes like me it is hard to choose a suitable bike from that list.

    If it wasn't for a trading post search for "learner" bikes I would never have known that a bike like the KLE500 was available as a learner bike.
  17. If you were looking at targeting it towards first bike buyers, the other issue with getting your information through datamining is that I could tell you how I ride, what my needs are and what bike I'm on, however my VTR is my first bike, I've never ridden another road bike and to be honest I'd have no idea if there was a bike out there better suited to how I ride. If you pick up information from a lot of newbies like me... maybe you wouldn't be getting the "best" bikes for each use... altho maybe others did more research and didn't fall over the first LAMS bike they came to drooling and begging for the keys :oops:
  18. Bikes always come down to opinions. But you are absolutely correct. That's why the question of "which is the best learner bike" should not be asked to you directly. The real skill of data mining is knowing which questions to ask. You would be able to provide reasonable information on how good a VTR was to a first rider such as yourself with your wants and needs (whatever they are/were).
  19. This sounds like such a great idea, although it will be a massive can of worms.

    The important part will be with your database definitions and classifications of motorcycle classes etc. as more than likely, the allocation of a bike to a certain category will be a subjective thing.

    Add too many bikes to too many sub-categories and the data is useless with a results list that is as long as your arm. Add too few, and people will be excluding bikes from their results list that could potentially be a perfect fit.

    How about a database that spits out the exact same result each time?
    Q. Which bike?
    A. Go and enjoy test riding them all.

    *giggles* ;)
  20. Yeah, I had imagined it with photos of the recommended bikes. After all, the advice around here, after covering the main bases of intended uses, is most often "choose the one you fall in love with". And what's a dating site without pix? ;)