IIT-JEE is an entrance examination that 18 year olds in India give for admission to the top engineering universities in the country. KVPY-SA is an examination that most/some of these students give 2 years earlier.

I was curious to find out how good of an indicator featuring on the KVPY ranklists is for eventual success in the JEE.

Results

• 60.0% (120/200) of students who cleared KVPY-SA in 2008 cleared JEE in 2010
• 74.9% (161/215) students who cleared KVPY-SA in 2009 cleared JEE in 2011
• 66.1% (156/236) students who cleared KVPY-SA in 2010 cleared JEE in 2012
• 62.9% (178/283) students who cleared KVPY-SA in 2011 cleared JEE in 2013
• 83.2% (467/561) students who cleared KVPY-SA in 2012 cleared JEE in 2014
• 67.0% (389/581) students who cleared KVPY-SA in 2013 cleared JEE in 2015

How does NTSE Fare?

NTSE is another examination that the students take in their 8th grade. I ran a similar analysis for this examination.

• 46.9% (350/747) students who cleared NTSE in 2010 cleared JEE in 2014
• 42.5% (329/774) students who cleared NTSE in 2011 cleared JEE in 2015

Average Ranks

KVPY

Of the students who cleared both KVPY and JEE, what is the average of the top 100 JEE Ranks.

• JEE(2010) and KVPY(2008) is 1151.3
• JEE(2011) and KVPY(2009) is 342.7
• JEE(2012) and KVPY(2010) is 273.1
• JEE(2013) and KVPY(2011) is 330.4
• JEE(2014) and KVPY(2012) is 89.1
• JEE(2015) and KVPY(2013) is 141.7

NTSE

• JEE(2014) and NTSE(2010) is 295.2
• JEE(2015) and NTSE(2011) is 439.5

What about those who cleared all of these exams?

• 133 cleared NTSE in 2010, KVPY in 2012, JEE in 2014. Average(n=100): 681.8
• 125 cleared NTSE in 2011, KVPY in 2013, JEE in 2015. Average(n=100): 1182.2

Not sure what to make of this. I was expecting the number to be higher?

What do these results suggest?

• Selection in KVPY is a stronger indicator than selection in the NTSE ( ~60% vs ~45%)

Other than that, I’m unsure if any strong inferences can be drawn from the results.

Points of Interest

• I think it’s cool how the numbers are largely consistent across multiple years.
• I’m surprised by the seemingly low 60% success rate by the KVPY. I’m not quite sure what explains why the other 40% did not eventually do well enough to even feature on the extended ranklist for the JEE. It shouldn’t have been that difficult if they managed to clear this 2 years ago. Since selection in the KVPY means that they have guaranteed admission into IISc, did they just not care enough anymore? I shouldn’t speculate. There are too many possibilities.
• The average rank of the top 100 students who cleared all three examinations being much higher than the other averages feels slightly strange but I’m not sure what to make of it.

Methodology

KVPY Data was available on the official website http://kvpy.iisc.ernet.in/main/fellows.htm

JEE Data was made available at https://captnemo.in/projects/iitjee/

NTSE Data was made available as PDFs on the NCERT website http://www.ncert.nic.in/

I extracted the data into CSVs, defined a common format to only include names, registration number and ranks and then ran a comparison for the equivalence of names.

I ran a fully-equivalent comparison so “Arvind Kumar” and “Kumar Arvind” would not be a match. This means that the actual values will possibly be off by a tiny percentage. I also did not run any duplication checks but again, this should not be cause a lot of variation.

Tools used

• Ruby to parse the KVPY HTMLs to CSVs and eventually to run the analyses
• Tabula to convert the NTSE PDFs to CSVs

Code

The data used in this project is made available at → (Link coming soon)

If you end up doing some further analysis on this, do let me know!

Credits

Thanks to Nemo and SDS Labs for making the JEE-data available that even enabled all of this analysis.