A few suggestions to improve:
* Consider that most users rank 1-10 to stack rank their own games. BGG explicilty encourages this, insisting it's YOUR rating and there is no wrong way of arriving at it. I'd try to normalize for this (i.e. convert user collections from ratings into stack rank percentiles, see where games end up on average).
* Board games objectively got better over the years. There is more material on mechanics, it's easier to publish than ever, simulating and playtesting games is easier, etc. When I look at some (not all!) old games on top 100 I always think "come on, it's a good game but has no place here". It would be interesting to see rankings based on ratings assigned in a specific year.
* Drop the notion of complexity as something to correct for. As others mentioned this is a very inconsistent data point. Personally I'd rather play two 8.5 games in an evening than one 9. Perhaps fun per hour would be better, but this is personal bias, no matter how you assign the weight you will get complaints :)
* Kids and party non-gamer games are almost a distinct universe from gamers games. There is no way to fix or account for that. It has nothing to do with compexity or play time. I think this is more down to a combination of how well the game was playtested combined with the distribution channels available to the publisher. Gamers will happily play 15 minutes gimmick games, non-gamers will pull out Monopoly in a pub and go throught this monstrous waste of cardboard while downing pints.