myLSN.info Forum

[LSATSCORES2012]

Hey folks, added a new feature that I thought I'd let you know about:

Now, on searches, there is a link to an image which you can embed in posts instead of linking to the page. (You can still link to the page if you'd like - just copy and paste the URL from your browser's URL bar instead).

These images are just normal images, but contain the relevant results for a range of schools. You can adjust the range you want displayed in the image by entering a new range in the fields below the image URL given (by default, the range is the T14.)

For example, below is the default image for this search:



And this is an image adjusted to be from school #50 to school #70:



And for HYS:



HYSCCN:



My hope is that this might make it easier to help other users, that way they don't have to visit a whole new website - you can just embed the information that you're using in your posts.

[LSATSCORES2012]

FYI I just integrated some upgrades that will help server performance - please let me know if you experience any errors so that I can fix 'em.

[helix23]

Thank you for all your work on this. I use the site a ton.

[Jaqen]

Very cool. Thanks!

[LSATSCORES2012]

Thanks for the support, everyone.

Just applied an update which should allow the website to sync with LSN every couple of hours without overwhelming either LSN's server or my server - if you experience any problems or notice anything not syncing properly, please let me know!

[LSATSCORES2012]

FYI any search/ranking URL's generated since 5AM yesterday (Sat.) will no longer work because I had to restore a database backup. Sorry for the trouble!

[LSATSCORES2012]

Thanks to whoever pointed out that ED/Time Submitted hasn't been working - should be fixed now, the problem was caused by an update I applied a few hours ago.

[LSATSCORES2012]

So, a few searches on mylsn will now return slightly different results (as of around 2PM this afternoon). There is a glitch on LSN where some user pages aren't actually displaying their highest LSAT score, but a score from a retake. I've made this manually identify their actual score (which is hidden in the page's code) rather than just taking the score that LSN displays. This also means that the score displayed on mylsn for a user might occasionally be different than what's shown on LSN, but the one on mylsn should be their actual highest score.

This problem only applies to 4% of LSN applications, so I don't think it should have a major effect on any searches, but it might impact those with small sample sizes.

[LSATSCORES2012]

Someone sent me feedback noting that if someone changes their LSN username then the same profile under the two different usernames is entered in the myLSN database. First, thank you for the feedback! Unfortunately I can't think of any good way to fix this during the live updates, which occur several times a day, but I do erase then completely re-sync the database for the current cycle with LSN every few weeks (I don't want to do it too often for fear of pulling on LSN's server resources), and during these syncs only the single username should show up. I'm fairly confident that the number of people who change their usernames on LSN (I actually didn't even know it was possible until I got the email with your message) is little enough that it shouldn't create any large errors in the time between syncs.

But thank you again for the email... I'm sorry that I can't fully fix the problem, but rest assured that it's not as large of an issue as it might initially have seemed.

Also, when you submit feedback I recommend leaving an email or username so that I can contact you directly rather than writing these public PSA's (but it's not required, so don't let that stop you from clicking the feedback button - even if it's just to write 'hi!')

[ManOfTheMinute]

Them pop-ups are getting annoying... is there no better way to display that information? Like a banner or something?

[LSATSCORES2012]

Them pop-ups are getting annoying... is there no better way to display that information? Like a banner or something?

Hmm... I'll take it down later today and think of alternatives.

[laxbrah420]

Someone sent me feedback noting that if someone changes their LSN username then the same profile under the two different usernames is entered in the myLSN database. First, thank you for the feedback! Unfortunately I can't think of any good way to fix this during the live updates, which occur several times a day, but I do erase then completely re-sync the database for the current cycle with LSN every few weeks (I don't want to do it too often for fear of pulling on LSN's server resources), and during these syncs only the single username should show up. I'm fairly confident that the number of people who change their usernames on LSN (I actually didn't even know it was possible until I got the email with your message) is little enough that it shouldn't create any large errors in the time between syncs.

But thank you again for the email... I'm sorry that I can't fully fix the problem, but rest assured that it's not as large of an issue as it might initially have seemed.

Also, when you submit feedback I recommend leaving an email or username so that I can contact you directly rather than writing these public PSA's (but it's not required, so don't let that stop you from clicking the feedback button - even if it's just to write 'hi!')

Your site is significant enough now that if they were going to ask you to cease/desist screen scraping, I think they already would have --do you think it's worth asking them for an API though? Or JSON/XML? Their site becomes more valuable the better the stats are, and you provide that to them for free

[LSATSCORES2012]

Someone sent me feedback noting that if someone changes their LSN username then the same profile under the two different usernames is entered in the myLSN database. First, thank you for the feedback! Unfortunately I can't think of any good way to fix this during the live updates, which occur several times a day, but I do erase then completely re-sync the database for the current cycle with LSN every few weeks (I don't want to do it too often for fear of pulling on LSN's server resources), and during these syncs only the single username should show up. I'm fairly confident that the number of people who change their usernames on LSN (I actually didn't even know it was possible until I got the email with your message) is little enough that it shouldn't create any large errors in the time between syncs.

But thank you again for the email... I'm sorry that I can't fully fix the problem, but rest assured that it's not as large of an issue as it might initially have seemed.

Also, when you submit feedback I recommend leaving an email or username so that I can contact you directly rather than writing these public PSA's (but it's not required, so don't let that stop you from clicking the feedback button - even if it's just to write 'hi!')

Your site is significant enough now that if they were going to ask you to cease/desist screen scraping, I think they already would have --do you think it's worth asking them for an API though? Or JSON/XML? Their site becomes more valuable the better the stats are, and you provide that to them for free

I hope you're right... at the very beginning, before I even launched the website, I asked them for permission to do it. They replied back and asked for a link to the website to check it out so I sent it over after I had launched it and I never got a reply. So they definitely know about it and it would suck if I had to bring it down. (If anyone is interested, I can send you a copy of this email discussion.)

An API would be nice because it would allow me (and others, too) to more efficiently access the information on LSN, so I think that's a great idea.

[laxbrah420]

Someone sent me feedback noting that if someone changes their LSN username then the same profile under the two different usernames is entered in the myLSN database. First, thank you for the feedback! Unfortunately I can't think of any good way to fix this during the live updates, which occur several times a day, but I do erase then completely re-sync the database for the current cycle with LSN every few weeks (I don't want to do it too often for fear of pulling on LSN's server resources), and during these syncs only the single username should show up. I'm fairly confident that the number of people who change their usernames on LSN (I actually didn't even know it was possible until I got the email with your message) is little enough that it shouldn't create any large errors in the time between syncs.

But thank you again for the email... I'm sorry that I can't fully fix the problem, but rest assured that it's not as large of an issue as it might initially have seemed.

Also, when you submit feedback I recommend leaving an email or username so that I can contact you directly rather than writing these public PSA's (but it's not required, so don't let that stop you from clicking the feedback button - even if it's just to write 'hi!')

Your site is significant enough now that if they were going to ask you to cease/desist screen scraping, I think they already would have --do you think it's worth asking them for an API though? Or JSON/XML? Their site becomes more valuable the better the stats are, and you provide that to them for free

I hope you're right... at the very beginning, before I even launched the website, I asked them for permission to do it. They replied back and asked for a link to the website to check it out so I sent it over after I had launched it and I never got a reply. So they definitely know about it and it would suck if I had to bring it down. (If anyone is interested, I can send you a copy of this email discussion.)

An API would be nice because it would allow me (and others, too) to more efficiently access the information on LSN, so I think that's a great idea.

It would be pretty retarded of them to get upset --and I get the feeling they're just good people like you. Honestly I'm surprised they haven't just bought you and your site.

Also I suck at coding, but to check for duplicate users, if you'd just make a set of gpa, lsat, [school + status] and compare, are you likely to get a lot of false positives? Not sure how easy it is in whatever language you're using, but set(x) == set(y) in python would do the trick

[LSATSCORES2012]

Someone sent me feedback noting that if someone changes their LSN username then the same profile under the two different usernames is entered in the myLSN database. First, thank you for the feedback! Unfortunately I can't think of any good way to fix this during the live updates, which occur several times a day, but I do erase then completely re-sync the database for the current cycle with LSN every few weeks (I don't want to do it too often for fear of pulling on LSN's server resources), and during these syncs only the single username should show up. I'm fairly confident that the number of people who change their usernames on LSN (I actually didn't even know it was possible until I got the email with your message) is little enough that it shouldn't create any large errors in the time between syncs.

But thank you again for the email... I'm sorry that I can't fully fix the problem, but rest assured that it's not as large of an issue as it might initially have seemed.

Also, when you submit feedback I recommend leaving an email or username so that I can contact you directly rather than writing these public PSA's (but it's not required, so don't let that stop you from clicking the feedback button - even if it's just to write 'hi!')

Your site is significant enough now that if they were going to ask you to cease/desist screen scraping, I think they already would have --do you think it's worth asking them for an API though? Or JSON/XML? Their site becomes more valuable the better the stats are, and you provide that to them for free

I hope you're right... at the very beginning, before I even launched the website, I asked them for permission to do it. They replied back and asked for a link to the website to check it out so I sent it over after I had launched it and I never got a reply. So they definitely know about it and it would suck if I had to bring it down. (If anyone is interested, I can send you a copy of this email discussion.)

An API would be nice because it would allow me (and others, too) to more efficiently access the information on LSN, so I think that's a great idea.

It would be pretty retarded of them to get upset --and I get the feeling they're just good people like you. Honestly I'm surprised they haven't just bought you and your site.

Also I suck at coding, but to check for duplicate users, if you'd just make a set of gpa, lsat, [school + status] and compare, are you likely to get a lot of false positives? Not sure how easy it is in whatever language you're using, but set(x) == set(y) in python would do the trick

So I have a copy of every user's full page in the DB (just in case I want to run an analysis, I don't have to download a whole new copy every single time). I could just search for duplicate page texts and have it delete the oldest copy... the pages wouldn't be identical, though, because of the change in username... and if they change anything else in the window between updates the program wouldn't detect them as updates... but it could probably catch most of them. That's a good idea, I think I'll go ahead and do that later today.

[LSATSCORES2012]

The site recently passed half a million visits (just in time for the imminent 1 year anniversary of its launch!), so I thought I'd post again how thankful I am to TLS for all of your support. Many of the ideas which made the website possible came from you, and, of course, nearly all of the data, being user generate, came from your profiles on LSN.

Here are some fun statistics, to celebrate:

Since about a month after the site started last July, there have been:

more than half a million pageviews...

in more than 100,000 visits...

by more than 30,000 unique visitors...



The average user stays on the website for about four minutes.

Nearly 97% of users speak English, and about 1% speak Korean. From there, the numbers drop below 1%.

92% of users are from the US, about 2% are from Canada, and about 1.5% from South Korea.

Within the US, the top 5 states in terms of numbers of visitors are: 1. California, 2. New York, 3. Florida, 4. Texas, 5. Illinois

About 30% of all visitors only ever visit the website once. Another 10% only visit it twice, and another 10% visit it from 3-8 times. 10% of users visit 9-25 times, and another 10% 26-100 times. About 10% of users visit 100-200 times, and 20% of users have visited the website more than 200 times.

The most frequently used browser is Chrome, then Safari, then Firefox. Internet Explorer has less than 10% of the browser share on mylsn.

About 50% of users have Windows, 40% have macs, 9% access the site from a mobile device, less than 1% access the site from Linux, and there have been 68 visits from Chrome OS.

The majority of visitors use comcast for their internet service, followed by Road Runner Holdco (what's that?).

More visits were registered to MyLSN from the University of Miami than from any other school in the country. (Admittedly, that's probably because I went to UM, and I needed to access it rather frequently for debugging purposes).

After UM, we have:

1. Columbia
2. Yale
3. Duke
4. NYU
5. Cedarville University (?)

233 visits have been registered from a New York City subway car.

The sources of traffic, ranked by the amount of traffic I get from them, are:

1. TLS
2. Direct (they just typed mylsn.info in their browser)
3. Google
4. Collegeconfidential
5. Reddit
6. Bing
7. Somethingawful.com

The most frequently accessed page is the search page, followed by the main page, then the timeline page, then the admissions graphs. The most frequently accessed search is http://mylsn.info/wfxuju (from a link graciously posted by john_rizzy_rawls here)

Chrome loads the website fastest, in an average of 1.53 seconds, followed by Firefox at 2.05, IE at 2.35, and the Android Browser at 4.63 seconds. For some reason I don't have info on Safari for this.

And that's all I got! I found it to be interesting - hope you did to. And, again, thanks for everything TLS!

[North]

Dude, thank you.

[LSATSCORES2012]

This is totally unrefined and not ready for production use, but I wanted to know if anyone has any input on this: http://www.mylsn.info/mychances.php

It's basically like the search, but you enter an LSAT score and GPA and your URM status and it automatically runs a search for you, adjusting the LSAT and GPA range so that it doesn't cross a median and trying to increase the sample size to at least 20 (without expanding the range too much). It then reports the level of confidence of the result, based upon how much it had to expand the range and the sample size.

If I were to follow through with this I would add more schools and remove all the data to the right, of course, but right now it's there for testing purposes.

Anyone have any thoughts or input? Thanks in advance!

[mvonh001]

I don't like how you distinguish the different confidence levels, It is confusing. for example if it says there is a low chance but the confidence is low also, does that mean i have a high chance ( i know it doesnt, but others may think that it does)

ETA you should also addd an ED option.

[LSATSCORES2012]

I don't like how you distinguish the different confidence levels, It is confusing. for example if it says there is a low chance but the confidence is low also, does that mean i have a high chance ( i know it doesnt, but others may think that it does)

ETA you should also addd an ED option.

Thanks for the input I added ED and fixed a small glitch in the code, and replaced the confidence words with +'s.

I get what you mean about confidence. Would using the term "robustness" make it better? Or does using the +'s like I have it now make it more comprehensible? Or, would, perhaps, a little bar that is small if there is low confidence and large if there is great confidence be better? Or do you think I should remove it altogether?

[mvonh001]

I don't like how you distinguish the different confidence levels, It is confusing. for example if it says there is a low chance but the confidence is low also, does that mean i have a high chance ( i know it doesnt, but others may think that it does)

ETA you should also addd an ED option.

Thanks for the input I added ED and fixed a small glitch in the code, and replaced the confidence words with +'s.

I get what you mean about confidence. Would using the term "robustness" make it better? Or does using the +'s like I have it now make it more comprehensible? Or, would, perhaps, a little bar that is small if there is low confidence and large if there is great confidence be better? Or do you think I should remove it altogether?

What are you using to determine confidence and what are your parameters for getting accepted?

ETA: i like the + / - infinitely better than before. btw

But after looking at it a little longer. If there is something with a ++ will there ever be something with a --? or a - at all? If you are not confident in the outcome of the result you should just leave it blank, and as you get more confident you should get higher -- like a percentage scale of how confident you are.

I just read your suggestion of having a bar... I think that could work better than the +++, for example if there is no confidence and you could be in or out, you put no bar, yet if you are very confident you put a long(/thick?) bar?

Or maybe get rid of confidence and color code everything. For example if you are certain about an outcome make it green, if you are uncertain yellow, and if you are not certain at all make it red.?? IDK im just spitballing here.

If you would like to chat i can create a chatroom and i can go through it with you in real time

[John_rizzy_rawls]

Great stuff, as always. The confidence predictor is interesting.

[homie1515]

You should not make it say insufficient sample if there are only 2 data points, you should just reduce confidence in the outcome.

[mvonh001]

You should not make it say insufficient sample if there are only 2 data points, you should just reduce confidence in the outcome.

^ Agreed.

[LSATSCORES2012]

As has been suggested, it will now display the chances as long as the sample size is greater than one and there is a sufficient confidence level.

The confidence is determined by a combination of: sample size and how much/whether the LSAT and GPA range need to be expanded in order to get a sufficient sample (which I changed to 7), extremeness of the percentage accepted, how well the results agree with (or conflict with) other ranges (for example, if a number has a low acceptance rate and applicants with better numbers also have low acceptance rates, this will increase the confidence of the unlikely prediction; if a number range has a low acceptance rate but people with lower numbers have a high acceptance rate this will lower the confidence).

ETA: and medians relative to LSAT score (that is, if someone is below both medians but it shows a high acceptance rate, confidence is lowered, whereas if they're above both medians and there is a low acceptance rate, confidence decreases)

Thanks again for the input. Any more input is definitely welcome