search

TaddyLab / hockey

Chicago Hockey Analytics
7 stars 2 forks source link

add results writeup to tex #5

Closed mataddy closed 9 years ago

mataddy commented 9 years ago

@sentian and I will fill in the application and results section using the goals, corsi, and fenwick analysis executed up through issue #4. A candidate layout would be

I won't be able to devote any time until mid-next week on this, so if @sentian can get a rough draft in there then I'll finish up and bring the paper on home.

mataddy commented 9 years ago

cool, thanks. and yes that is what I was trying (unsuccessfully) to say above: there's a difference between not having enough data and not accounting for the amount of ice time. of course we could separate these things by looking at the posterior for the ppm; but our current point estimation tools conflate the tools. it is an argument for bayes!

rbgramacy commented 9 years ago

Ok, I put a little note in the discussion section. Thanks for being patient with me. -B

On Tue, Sep 29, 2015 at 8:55 PM mataddy notifications@github.com wrote:

cool, thanks. and yes that is what I was trying (unsuccessfully) to say above: there's a difference between not having enough data and not accounting for the amount of ice time. of course we could separate these things by looking at the posterior for the ppm; but our current point estimation tools conflate the tools. it is an argument for bayes!

— Reply to this email directly or view it on GitHub https://github.com/TaddyLab/hockey/issues/5#issuecomment-144231128.

sentian commented 9 years ago

@mataddy

  1. We don't have data for 2004-2005 lockout. Probably vertical line for 2005-2006?
  2. Are we doing it for goals, corsi, or fenwick? The way I did figure 1.4 is that the white area is for players having zero ppm in either goals or corsi, then fill in other players in the grey areas.
mataddy commented 9 years ago

1:yes, makes sense. 4: ah, I see. I'd prefer separate plots: one plot for goals and one for corsi.

sentian commented 9 years ago

I've tasks 1,2,4, and will do 3 tmr morning. I've uploaded a new R file called todo.R, which is the code for the tasks in the to-do list. Just in case you still want to do adjustment of the figures. It's basically self-contained and I've labeled them which you will find pretty easy to play with. We can delete it after finishing everything since it's duplicated with the code I've uploaded.

Note: everything I've done is without playoff effects. I rerun models by removing the playoff effects. And all the statistics with '.avg' are corresponding to the seasonal performance.

I will be almost free tmr. So feel free to interrupt me for any further works.

1&2: figures 1.3, 1.4 and 1.5. I leave the range of their y-axis the same for easier comparison. But it's fairly easy to change (just change ymin, ymax and possibly x.legend & y.legend).

4: figure 1.6 (season 1314), figure 1.7 (all seasons)

mataddy commented 9 years ago

thanks. why no playoff effects? one of the reason we argue corsi stats might be useful is that they allow us to separate playoff and reg performance. On Sep 30, 2015 2:25 AM, "Sen Tian" notifications@github.com wrote:

I've tasks 1,2,4, and will do 3 tmr morning. I've uploaded a new R file called todo.R, which is the code for the tasks in the to-do list. Just in case you still want to do adjustment of the figures. It's basically self-contained and I've labeled them which you will find pretty easy to play with. We can delete it after finishing everything since it's duplicated with the code I've uploaded.

Note: everything I've done is without playoff effects. I rerun models by removing the playoff effects. And all the statistics with '.avg' are corresponding to the seasonal performance.

I will be almost free tmr. So feel free to interrupt me for any further works.

1&2: figures 1.3, 1.4 and 1.5. I leave the range of their y-axis the same for easier comparison. But it's fairly easy to change (just change ymin, ymax and possibly x.legend & y.legend).

4: figure 1.6 (season 1314), figure 1.7 (all seasons)

— Reply to this email directly or view it on GitHub https://github.com/TaddyLab/hockey/issues/5#issuecomment-144312646.

mataddy commented 9 years ago

@sentian I just looked at the new pictures and they look awesome! Figure 1.3 alone (correlation between salary and PPM) should have NHL teams lining up to talk to us ;-)

For task 3 (and related to my comment above) we definitely want this table to be for model that included playoff effects, because I want to discuss how corsi allows us to differentiate between playoff and regular searon. We only need for corsi, not for fenwick. Feel free to give me a call if you want to chat (or if you need me to run some of this stuff). I'm booked 1030-11 central but otherwise free.

sentian commented 9 years ago

@mataddy Plots with 'regular' in the caption now means plots for regular session stats, while 'season' means seasonal stats. But they can be hardly distinguished as the playoff does not contribute much. So I think we just delete all the 'season' plots?

Also, Corsi table is checked in.

The only task left is the under/over paid table. I need to hold a TA session until 2pm EST. I will do it by then.

And 'Figure 1.3 alone (correlation between salary and PPM) should have NHL teams lining up to talk to us', I did not quite get this, more specific?

mataddy commented 9 years ago

great!

For the NHL teams comment: I just meant that we are able to show our metrics are more correlated with salary will have them interested.

mataddy commented 9 years ago

hi @sentian following up on our call earlier; do you have an ETA on this stuff? Let me know if you need me to pick off any of the tasks. I'll start finishing up the draft in about an hour.

sentian commented 9 years ago

@mataddy Can you have a look at the tables 1.3 and 1.4? And once you have a chance, give me a call at 312-888-5361.

mataddy commented 9 years ago

yes I will in one min. however it seems that your check-in overwrote bobby's last changes and his updates to PPM; can you fix?

On Wed, Sep 30, 2015 at 3:32 PM, Sen Tian notifications@github.com wrote:

@mataddy https://github.com/mataddy Can you have a look at the tables 1.3 and 1.4? And once you have a chance, give me a call at 312-888-5361.

— Reply to this email directly or view it on GitHub https://github.com/TaddyLab/hockey/issues/5#issuecomment-144534580.

mataddy commented 9 years ago

@rbgramacy fyi I doubt that this will be finished until late tonight; just too much writing still left to do. you can either wait up for it to submit or (better I think) we're a day late and go over a final version together in person tomorrow.

mataddy commented 9 years ago

everything is now done except the salary section. all plots are in I just need to spin a yarn.

rbgramacy commented 9 years ago

Cool. We can send it in by the end of the week -- I just gave Tim the heads up. (I blamed the dog for eating some of our plots.) I can do some editing on the plane tomorrow morning if you let me know what sections are ready.

-B

On Wed, Sep 30, 2015 at 7:53 PM mataddy notifications@github.com wrote:

everything is now done except the salary section. all plots are in I just need to spin a yarn.

— Reply to this email directly or view it on GitHub https://github.com/TaddyLab/hockey/issues/5#issuecomment-144576861.

mataddy commented 9 years ago

@rbgramacy the paper is done from my end; if you give it a read on the plane we can chat tomorrow morning.