Go Back   Team-BHP > BHP India > The Indian Car Scene


Reply
 
Thread Tools Search this Thread
Old 1st June 2010, 11:36   #241
Senior - BHPian
 
SkyWalker's Avatar
 
Join Date: Jul 2006
Location: Bangalore
Posts: 1,531
Thanked: 122 Times
Default

Echo - excellent effort -thanks.
The future not looking too bright for petrols. There only one petrol (ANHC) in the top 5 list. Toyota is reacting, hope other petrol only manufacturers wake up.
SkyWalker is online now   Reply With Quote
Old 1st June 2010, 12:30   #242
BHPian
 
echo77's Avatar
 
Join Date: May 2010
Location: goa
Posts: 325
Thanked: 33 Times
Default

Fiat prices being updated?

Fiat India shows:
Quote:
This page is under updation. Kindly revisit later.
The price link in Linea :: does not work as well.
echo77 is offline   Reply With Quote
Old 1st June 2010, 12:33   #243
BHPian
 
echo77's Avatar
 
Join Date: May 2010
Location: goa
Posts: 325
Thanked: 33 Times
Default

Quote:
Originally Posted by SkyWalker View Post
Echo - excellent effort -thanks.
The future not looking too bright for petrols. There only one petrol (ANHC) in the top 5 list. Toyota is reacting, hope other petrol only manufacturers wake up.
Thanks, Skywalker.

Jun/7 is the next govt decision on fuel price hikes. The question is if it is going to be another ad-hoc hike or are prices going to be deregulated (at least partially)

Some hints are that petrol may get deregulated, but it will be difficult to deregulate diesel because of inflation and all truckers going on strike. Petrol could shoot up to about 55 in Delhi... not sure about diesel, though.
echo77 is offline   Reply With Quote
Old 1st June 2010, 12:48   #244
Distinguished - BHPian
 
smartcat's Avatar
 
Join Date: Jun 2007
Location: Bangalore
Posts: 2,970
Thanked: 11,907 Times
Default

echo77, excellent stuff. The normalized rankings is more logical and makes lot of sense. The diesels take a huge lead when you consider the massive difference in the running costs of petrol & diesel.
smartcat is online now   Reply With Quote
Old 1st June 2010, 13:40   #245
BHPian
 
echo77's Avatar
 
Join Date: May 2010
Location: goa
Posts: 325
Thanked: 33 Times
Default

Thanks to you smartcat! It was only possible because of your sheet
echo77 is offline   Reply With Quote
Old 1st June 2010, 13:47   #246
ACM
Distinguished - BHPian
 
ACM's Avatar
 
Join Date: Mar 2008
Location: Mumbai
Posts: 4,152
Thanked: 2,456 Times
Default

Value retention figures will obviously not be available for 5 years and using the ones for 2 years makes absolutely no sense, as it is unrealistic and not gradual as in the case of 5 years. So suggest that since this data is not available it should not be used. Also it takes a longer bit of time for the true value (retention) to be ascertained. For example today Dzire may have high value retention (perceived) but if Linea takes off after introduction of the new engine the it's value would increase over a 5 year period, also if Manza overtakes Dzire in sales in the near future (very likely - considering the sales in the last few months) then Manza will end up having higher value retention compared to the Dzire in the longer run.

Dealer inputs are subjective and coupling them with carwale inputs is not fair. So suggest that this parameter be removied as it is baised and cannot be documented (all that echo and Smartcat) have attempted in this thread gets negated by adding such a parameter.

Also as far as car sales are concerned, this is also a function of cost of the car (though that is not the only factor). It is in some ways the sales of a car that smartcats final results would indicate in some ways for the long run. But using sales of a car itself as a parmeter too is not fair. it though objective is something that is skewed for new cars (too high or too low). Hence suggest that input too not be used.

Hence Sorry Acid - but am not in favour of using either of the parameters that you have used as they are not truly objective nor should they impact a car purchase.

One could then add brand ratings, Service ratings etc, but these too are not objective so too cannot be added in an objective study as mentioned at the start of the thread.
ACM is offline   Reply With Quote
Old 1st June 2010, 13:55   #247
BHPian
 
downsouth's Avatar
 
Join Date: Nov 2008
Location: 0.1933333, 32.58333
Posts: 357
Thanked: 23 Times
Default

Awesome job echo, my eyes were on the Linea since I drive one. I was surprised with the ratings for Linea 1.4 petrol since this sedan had so many features and did lose out in other areas, but the analysis has been practical. I will be watching this thread for more inputs.
downsouth is offline   Reply With Quote
Old 1st June 2010, 14:13   #248
BHPian
 
echo77's Avatar
 
Join Date: May 2010
Location: goa
Posts: 325
Thanked: 33 Times
Default

Thanks! My eyes on the Linea too. The petrol ratings are a bit disappointing, though. I will need another set of longish test drives of both fire and mjd.
echo77 is offline   Reply With Quote
Old 1st June 2010, 15:11   #249
BANNED
 
Join Date: Oct 2009
Location: Lucknow/Gurgaon
Posts: 1,022
Thanked: 91 Times
Default

Quote:
Originally Posted by smartcat View Post
great work SMARTCAT, rated 5 star to your thread, a detailed and very nice observation and comments, you have covered even the minutest of the details, and according to this my fiesta 1.6SXi stands at three.
ruchirtnt is offline   Reply With Quote
Old 1st June 2010, 16:19   #250
BHPian
 
Join Date: Jan 2010
Location: Chennai
Posts: 393
Thanked: 110 Times
Default

Quote:
Originally Posted by ACM View Post

Hence Sorry Acid - but am not in favour of using either of the parameters that you have used as they are not truly objective nor should they impact a car purchase.

One could then add brand ratings, Service ratings etc, but these too are not objective so too cannot be added in an objective study as mentioned at the start of the thread.
Er, Lets jus agree to just look at this analysis from an light and academic perspective and not get too passionate about it.

Any data, objective or subjective, will never be able to quantify everything about a car. I totally agree that any recorded data has its own internal limitations, and on top of that even an analysis on perfect data has its own limitations. So this excercise will always have plenty of holes from a logic perspective

I had a posted a disclaimer on the very idea of using these values in the first post itself.
acidkill is offline   Reply With Quote
Old 6th June 2010, 13:50   #251
BHPian
 
Dr.AD's Avatar
 
Join Date: Apr 2010
Location: Bangalore/Pune
Posts: 428
Thanked: 1,091 Times
Default Scores by Compounding Method

Hello everybody,

Finally, I have done some (little) work as per the method I suggested earlier (Post # 48 in this thread). The results are presented in this post.

First of all, many thanks and all credit to smartcat for the data and all the initial hard-work. I have just taken the data and modified it only slightly.

Next, many thanks to echo77 for his work and his data as well. His work greatly simplified my work (because he already implemented the normalized system), and again I have taken his data and tweaked it only just. Credit to echo77 for the data used here.

Before, I present the results, I would like to elaborate on the differences in my analysis compared to previous analysis posted by smartcat and echo77.

1. Normalized rating within each category: As I explained in Post # 48, this would take care of the actual value (or score) in that category and not just the relative position. As it happened, echo77's normalization is exactly same as what I had proposed, and he has already implemented this. So I have just taken echo77's numbers here, with just one minor change: I have scaled the normalized scores to '0 to 1' scale (instead of 0 to 100 used by echo77). The reasons are related to special property of number 1, which is utilized in compounding the scores, as explained next.

2. Compounding the scores: Previously, both smartcat and echo77 used "total of all scores" (i.e. simple additions) as a way of obtaining the total score. I had argued against this, and I had suggested "multiplication" method. Let me explain the rational here:

The total product is only ad good as its 'weakest link'. In other words, if a car scored really pathetically in any one category, then it is a bad car regardless of how well it does in another category. "Addition" has a property of balancing out these scores across categories, which means that the total score is not a reflection of the weakest score, but more of an aggregate of everything. On the other hand, "multiplication" will be a true reflection of the 'weakest link', because a really bad score in any one category will pull down the multiplication heavily.

Therefore, I have used net score (called Compounded Score) as a multiplication of all scores. Let me explain the intuition behind this with some examples:

Let us consider 0 to 1 scale now. Multiplication of any numbers between 0 to 1 results into a number between 0 to 1. Thus, even the final number is nicely scaled and bounded, regardless of the number of categories.

If a car scores 1 in all categories (i.e. it is the best in all categories), all multiplications will result in net score of 1. Thus, net score of 1 indicates the best possible result. Now, let us consider only 3 categories for sake of simplicity. If a car scores 0.9 in each category, then we can say that we are losing out slightly in each category. But losing out slightly three times (in all three categories) results in overall a significant loss. The additive score (as used earlier) will result in score of 2.7 out of 3, which is same as 0.9 out of 1. Thus, addition does not penalize the car for being slightly worse in all three categories. However, now the Compounded Score (multiplication 0.9x0.9x.09) is only 0.729.Thus, this score is worse than 0.9 out of 1 given by additive method. This is the penalty for being slightly worse, but across all categories. The argument is that if I am slightly worse is everything, then in total I am more than slightly worse. In order to get the net result of 0.9, the car needs to score 1, 1, and 0.9 in the three categories, which really means that I am only slightly worse in only 1 category.

Now let us consider two cars: Car A scores 1, 1, and 0.4, whereas Car B scores 0.8, 0.8 and 0.8. The additive method will give same scores (2.4 out of 3, which is same as 0.8 out of 1) for both these cars. In reality, Car B is an all-rounder whereas Car A is hopeless in third category and it should be penalized for being so poor. The Compounded scores are 0.512 (0.8x0.8x0.8) and 0.4 (1x1x0.4) for Car B and Car A respectively. Thus, the Compounded Score penalizes Car A for having the worst 'weakest link' of the two cars, whereas additive scores are blind to this.

Further, if any car comes last in any category (i.e. normalized score of 0), then the Compounded Score will be 0 regardless of the scores in other category. This says that if the car is so bad in any one category, it is a bad car and such hopeless score can't be compensated by anything else. Basically as I explained in my earlier post, multiplication asks for 'this AND that' score, whereas addition asks for 'this OR that' score.

3. How to handle the OTR price? In earlier work, the OTR price was included as one of the categories. Then some comments were made that the cars that are in the top half of the results are good VFM and vice versa for the vars in the bottom half. I do not agree with this. There is a statistical flaw in this argument: If the OTR price is already included, then you can't again say which one is VFM. The effect of price (which is primarily related to the notion of VFM) is mixed together with the effects of everything else, and it is impossible to pinpoint in the end as to which car costs least for a given performance. Statistically, this phenomenon is called "Confounding of Effects". For example, a car in the top half could be very expensive, but still there because of great scores on everything else. But a car at the bottom can cost half of that, and can have lesser scores on everything else, and hence at the bottom. Now you can't say that the car at the top is more VFM. Both cars may cost the same if you consider price per unit score. Therefore, in order to comment on which car is VFM, we need to calculate total score per unit OTR price.

There is another motivation here. If the proposed Compounded Score, the car that is most expensive will get a normalized score of 0 for OTR category, and its total score will be 0 regardless of other scores. This is one of the side-effects of the proposed method (I never claimed it is free of defects ). However, this can be easily eliminated by removing OTR from the calculations of net scores and then considering scores per unit OTR, which anyway makes more sense as explained above.

Results:

Now, here are the result tables:

Table 1: Here, I have included OTR as one of the factors. Basically I have taken the data given by echo77 and just replaced "additions" by "multiplications".

Sedans Under Rs. 12 Lacs - A Quantitative Ranking-compoundedscores.jpg

As expected, you can see some changes, and more importantly, many cars given net score of 0. Now, there is one anomaly: Optra Diesel given net score of 0 for being the most expensive car (as explained above). This error is eliminated in Table 2.

Table 2: Here, I have eliminated OTR as one of the factors, but used it later to calculate score per unit OTR.

Sedans Under Rs. 12 Lacs - A Quantitative Ranking-scoresperunitotr.jpg

Now you can see that Optra Diesel is at the top.

Some interesting changes:

1. Optra Petrol has jumped up to 12 from 21. The reason is that it is about average on all categories, and not really poor in any one category. The multiplication favors this. For example, 0.5x0.5 is better than 0.8x0.2. Lack of any particular 'weakest link' is rewarded here for Optra.

2. Ford Fiesta (both petrol and diesel) moves down significantly. The reason being 'practicality', which is a real 'weak link' for Fiestas, and the rating system penalized it for that.

3. Numbers 20 through 26 are all bottom rankers, pulled down there by a score of 0 in one of the categories. However, these are more or less the same cars that were at the bottom earlier, and no real surprises here.

Disclaimers:

1. All data is used as given by echo77. I did not change any data whatsoever, except for the changes outlined above.

2. I do not claim that these rankings make sense in all cases. I am sure one can find defects and some cases which appear counter-intuitive. The purpose of this post is to present a different method and explain the statistical intuition behind it. The purpose is not to move any particular car up or down the table. The results are just coincidental to the method used.

Credits:

Once again, many thanks to smartcat and echo77 for the data and all the hard-work.

Any comments, suggestions and criticisms are welcome.

Thanks,
-AD
Dr.AD is offline   Reply With Quote
Old 6th June 2010, 18:38   #252
BANNED
 
Join Date: Nov 2008
Location: New Delhi
Posts: 690
Thanked: 152 Times
Infractions: 0/1 (5)
Default

Now all i wish if someone can take the pains to calculate service cost of each vehicle till say a lakh kms(Engine Oil + Air Filter + Oil Filter + Labour + Timing Belt + Clutch parts as per respective manufacturer service book) , tyre replacement at 50000 kms and total of few items like bonnet , bumpers , headlights (basically accidental items) and add on to this and it will be a dream comparison.

Last edited by Trust_In_Thrust : 6th June 2010 at 18:39.
Trust_In_Thrust is offline   Reply With Quote
Old 6th June 2010, 19:04   #253
BHPian
 
echo77's Avatar
 
Join Date: May 2010
Location: goa
Posts: 325
Thanked: 33 Times
Default

Excellent analysis, Dr.AD!

This brings up the above average performers in most categories.

Interesting to note that the Optra 2.0 D is more bang for the buck than the Manza, which is about 4.4L cheaper.
echo77 is offline   Reply With Quote
Old 6th June 2010, 20:02   #254
BHPian
 
Dr.AD's Avatar
 
Join Date: Apr 2010
Location: Bangalore/Pune
Posts: 428
Thanked: 1,091 Times
Default

Quote:
Originally Posted by echo77 View Post
Excellent analysis, Dr.AD!
Thanks, but as I already mentioned, all credit to smarcat and you for the data and also majority of the work.

Quote:
Originally Posted by echo77 View Post
This brings up the above average performers in most categories.
Exactly! This is the whole point of the multiplication idea. If a car is above average in all categories, it will score good in the net score. Then, if the car is above (or close to) average in all categories and very good in a few categories, it will score great in the net score. Finally, any car that is really bad (considerably below average) in any one category will score bad in the net regardless of how it does in other categories. These properties are achieved by: 1. Using multiplications and 2: the scores for each categories being in the range of 0 to 1.

Basically the idea is to look for overall good car without any serious flaws.

Quote:
Originally Posted by echo77 View Post
Interesting to note that the Optra 2.0 D is more bang for the buck than the Manza, which is about 4.4L cheaper.
Yup, and this is the reason I propose to take out the OTR factor from score calculations and then simply divide by OTR to get "score per unit OTR". This I believe is the true indication of VFM. In this particular case, Optra 2.0 D has a net score of 0.1614 for 12.2L OTR, whereas Manza Diesel has a score of "only" 0.0822 (almost half of Optra D) for OTR of 7.83 L. Thus, in Optra Deisel, one gets almost twice the score than Manza for about 1.5 times the cost, and hence Optra Diesel is more VFM than Mazna Diesel. This simple fact gets "confounded" with other factors when we include OTR in the score calculation (as was done previously).

By the way, this is totally OT, but confounding is a very interesting (and often complex) phenomenon occurring in a variety of engineering experiments done, in all fields of engineering design and analysis. However, it should be noted that in the context of the current analysis, I am using the term confounding very loosely (and vaguely), and the precise definition and meaning is different than that intended here.

Thanks,
-AD
Dr.AD is offline   Reply With Quote
Old 6th June 2010, 20:41   #255
BHPian
 
NutsNBolts's Avatar
 
Join Date: Jul 2009
Location: Kochi
Posts: 796
Thanked: 20 Times
Default

Dr AD, thanks for such a wonderful exercise and explanation. Of course, without the initial spadework by smartcat and echo77, it wouldn't have been achievable. My sincere thanks three of you and a request to mods to make this thread a sticky. This thread doesn't deserve to go un-noticed by people deciding to buy sedans in near future.
NutsNBolts is offline   Reply With Quote
Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
Ranking of hatchbacks in India Based on looks. vishnurp99 The Indian Car Scene 239 30th November 2016 02:33
Ranking of indian cities according to cars avishar Shifting gears 4 3rd December 2014 16:49
The Most Dependable Car Brands - Ranking Survey in US shenoytech The International Automotive Scene 15 22nd March 2011 15:35
Small Car Ranking Guidance Archon The Indian Car Scene 22 1st December 2010 10:44


All times are GMT +5.5. The time now is 19:58.

Copyright 2000 - 2017, Team-BHP.com
Proudly powered by E2E Networks