Quote:
Originally Posted by smartcat
Although it seems to be unfair, as you have mentioned yourself, the high & low values will cancel itself out - I think it is a good property! When we look at the overall scores via addition, the "normalization" will be done automatically. |
No, cancellation is not a good property. For example, if a car has very bad score on performance, then what use is a very good score on safety (or vice versa)? In your method, a great performance in one case will help the car balance out really pathetic performance in another case, due to the cancellation property. In reality, this is not true. A great performance is only worth it if everything else is at least at some decent level (and vice versa). Therefore, I am suggestion the "compounding" approach instead of the simple addition.
Simple addition has the property of presenting "OR" operation (i.e. I want to have performance OR drivability OR safety etc.) case whereas compounding (by multiplication) is just "AND" operation (i.e. I want to have performance AND drivability AND safety etc.).
Quote:
Originally Posted by smartcat Basically, multiple parameters & addition of points in each parameters automatically takes care of normalization, and also keeps the scoring system fairly simple. |
No, it does not.
Normalization takes care of fairness of scoring within a given parameter, and it has nothing to do with cancelation or otherwise over multiple parameters (the latter being taken care of by compounding mentioned earlier).
Somebody had earlier mentioned a good example: Suppose one car lags another in 0-100 time by 0.01s, then this car should not be one full point behind the other (this would be around 4% penalty if they are ranked number 26 and number 25 respectively). If we use normalization, the second car will get very low percentage of penalty, proportional to the 0.01% difference.
This has nothing to do with what happens on other parameters. Quote:
Originally Posted by smartcat Also this scoring system cannot be using to rank EQUIPMENT & SAFETY. |
Well, it can be. Let us say we give 1 point each for having critical safety equipment such as airbags, ABS, rear disc brakes etc. Then all cars which have all of these will get 1 point on normalized scale, all cars which have none of these will get 0, and any car which has a few of these will get score in-between 0 and 1, depending on how many equipments it has.
Quote:
Originally Posted by smartcat Doc, Are you a PHD type of doc? |
Unfortunately, yes :-)
Quote:
Originally Posted by smartcat As ACM suggested, I will stick to points system in descending order for simplicity, readability & usability. |
Completely fair! You have done a great analysis and I agree with your KISS principle. Again, we all appreciate your efforts, and I just wanted to make some suggestions to make it more accurate.
Quote:
Originally Posted by smartcat But I will also upload my XL sheet at the end - anybody can work on it with your much accurate system of ranking cars. |
Thanks, that will be great. I will try to do something with the sheet.
Quote:
Originally Posted by ACM But the normalization method could also be carried out in parallel. There is a major value to it as well.
I may be best if you could do an alternative calculation based on your system of normallization and post back to this thread the calculations with the normalization method. |
Great suggestion, but basically you are asking me to stop being lazy and do some real work
Now that is not easy :-) But certainly I will try my best as and when I get time.
By the way, Smartcat, please do not misunderstand me. I again say that you have done a phenomenal job and I sincerely appreciate your analysis and efforts. I did not mean to criticize it in any way, but just wanted to share some thoughts on what we call "scope for further improvement".