Campbell and I finally feel like we have some very productive plans on the horizon. This Thursday, Campbell and I will be meeting with Dr. Groves to do some very productive modeling. We are very excited to have this opportunity because we have been doing so much prep work for actual modeling, and while we've been able to do some on our own, Dr. Groves' personal help will certainly propel us to exactly where we need to be. Also, we are planning our winter meeting with Mr. Starr to reveal the strides that we've made over the course of the past few months. We plan to probably meet this Friday. Otherwise, our plan over break will be to collect more stats for a new draft class. This way, we will take our original model from our work this week with Dr. Groves, and then strengthen it with more and more statistics. We are very much on par to have a fully functional model by mid-February so that we can start preparations for our podcasts. Another interesting note is that the current college football season is winding down. With Bryce Young winning the Heisman trophy and multiple edge rushers that seem to be generational talents, we are thrilled to begin making our own mock drafts for this upcoming draft class as we continue to watch how the NFL standings play out in the winding weeks of the regular season.
1 Comment
December is already upon us. This year has truly flown by. We have had a few slow weeks due to Campbell's concussion and JP's college applications. However, we still feel confident about our progress going forward, as we have set up a meeting with Dr. Groves in person for next Tuesday. We have also decided to go in a different direction with our modeling software, due to the advice of Dr. Groves. She has recommended that we utilize Stata 17 Statistics Software, so that she can help us with the modeling. Additionally, we have been working through the tedious process of collecting defensive players college statistics one player at a time, because we were unable to find a collection of statistics that could be exported to Microsoft Excel. We believe that we will be able the have a productive month, with winter break in the near future.
Halfway through November and this year is moving faster than we would have expected. Still, Campbell and I are working hard and making strong progress on our project. Truthfully, this week specifically has been slow, as Campbell is essentially out of work with a concussion. We've spoken about plans moving forward, and are confident that it will just be a slight roadblock. Still, we are continuing to experiment with modeling and using our 2012 draft class to find equations. There is not a whole lot to report, as the process is pretty much confined to our modeling program, but we are excited to continue our project. We are also planning to meet Dr. Groves in person very soon and we may collect another batch of data from another draft class over Thanksgiving break in order to refine our model. Although we still have many months ahead of us, we are also beginning to preliminarily discuss what we will need for our podcast so that we can hit the ground running when the time comes. Otherwise, our project in its current state remains to be chipped away at, bit by bit every day, and we are excited to reach our final product within the next few months.
It feels good to finally make some progress back on our Fellows project. As our respective sports seasons are coming to an end and the first college application deadline is finally in sight, we decided to return to a real sense of Fellows dedication. First, we set up a third meeting with Dr. Groves. After a bit of miscommunication, we were able to connect with her to get jumpstarted into the next portion of our project. She came with ideas about how to realign our data for modeling and some information about usage on Stata. With that in mind, we have begun to consider moving from IBM SPSS to Stata for our modeling. Dr. Groves is more familiar with it, but also it has some capabilities that we did not expect to need, and learning its functions may be slightly simpler. After running some preliminary models with Dr. Groves, we also realized that the project had just gotten much easier. The reason for this change is because we we had just learned about linear regression modeling in our statistics class. Our Fellows project is, at its roots, linear regression. It requires some more complicated principles, but having sufficient background with linear regression makes our project much clearer. Heading into November, we are very confident in our progress. We are planning on meeting with Dr. Groves more often and hopefully meet her in person very soon to continue our modeling work.
As we continue to make our way through the busy fall season, we continue to make slow progress on the project. With a busy spirit week and a great deal of classwork it has been difficult to find time for the project. With college application deadlines coming up we anticipate more of the same in the next few weeks. Fortunately, we expected to be very busy in the fall, so we were able to get a large portion of the data collection done over the summer. Despite our bustling schedules, we continue to put in much time and effort into data collection and have set a meeting with Dr. Groves for the upcoming week. In this meeting, we hope to start moving our data from Excel into SPSS during this meeting, which is a very exciting step. We are extremely eager to get our data into SPSS so that we can begin the modeling process in the coming month.
Finally, as September winds to a busy end, we are proud to be titled as official "Severn Fellows". After an unexpected week-long sickness and a twice-presented presentation, we are able to leave summer behind and move onto our next steps. As mentioned in the previous blog post, our work was put slightly on hold with an incredibly busy school year so far and a lot of waiting for our presentation. The bus came to a screeching halt as Campbell came down with a sickness, but now fully recovered, we're incredibly excited to get back to work. The past few weeks have consisted mostly of some background thinking about how to get over our expected roadblocks and a bit of research about statistics and modeling, but otherwise, I'm willing to admit that progress has slowed (again, just due to the unforeseen circumstances; we're still deeply committed to our project). This has gifted us with both a slightly more urgent attitude and an incredible itch to get back to our project. In the coming months, as outlined in our fall plan, we are excited to finally do the modeling and are ready and willing to meet every obstacle that comes our way.
The past few weeks have been dedicated primarily to the return to school and immersing ourselves into our classes - certainly a busy couple weeks. As for Fellows, we have had one of our less productive periods. We have been working hard on our presentation and practicing plenty for our meeting next week. Otherwise, a lot of our effort has been put into a bit more research. After a few roadblocks that we discussed with Dr. Groves in our previous meeting, Campbell and I have tried our hardest to research articles and other ways to move over the roadblocks, while Dr. Groves has been talking with her own colleagues about creative solutions as well. We have done slightly more statistical gathering, but are mostly just doing some research to prepare for our modeling, which will occur shortly after our (hopeful) continuation onto a full Severn Fellow.
Oh how quickly the summer has flown by! It seems as if just yesterday we were meeting for the first time to discuss the fellows project. In the past two weeks we have continued to add to the vast collection of statistics on the players from the 2012 draft class. Along with combine statistics we have continued by adding numerous college statistics from the 2011 and 2010 college football seasons to our data. Additionally, we met with Dr. Groves to discuss how to go about making our model using SPSS. Whilst in our meeting we discussed how to fix a few problems we are having with the data collection. Specifically, we are looking for a way to speed up data collection for defensive players, as we have so many variables to collect. We also discussed the idea of creating a "dummy variable" that would help to balance out the players statistics based on their level of competition. Below we have attached images of our data collection in excel. The first image is a zoomed in image showing some of the statistics that we are currently collecting. The second image is zoomed out, showing the vast number of data points that we are collecting for each individual player. Although we zoomed out as far as we could within reason, we were only able to show the first 120 players of the 300 that we have collected data on. We continue to chug along with data collection and plan to meet with Dr. Groves again next week to discuss the best plan of action moving forward.
Two weeks to school! Hard to believe that summer is almost over, but we are excited to move into the daily groove of school as seniors. Frankly, there is not too much updating to be done at this point in our project. Over the past two weeks, we've completed lots of laborious data entry for the 2012 draft class. We began with combine stats for over 300 players, and once we gained access to the Pro Football Focus elite subscription, we were able to fill in their average qualifying season grade, which was taken by averaging the top five (qualifying, by our own criteria) season grades of any given NFL player. This average grade is the metric that we will be using to judge the success of a player at the peak of their career. From here, we plan to fill in some college statistics over the next two weeks, as well as reaching out to Dr. Groves again to check in on our progress and to see if she has any tips moving forward. For now, though, we are right on track from our original summer 40% plans, and we are looking forward to a busy fall.
As we head into August, we are pretty confident in our progress. Per Dr. Groves' advice, we've moved over to Microsoft Excel for the time being. While Excel may not be able to run the type of statistical analysis that we'll need, it is far superior for collecting data. Luckily it will be easy to transfer our data sheets to SPSS when necessary. Admittedly, we were slightly ambitious to try and collect data and immediately run analysis on it. Dr. Groves suggested that statisticians should always start with as much data as possible, and then decide from there which data is useful or important to our model. With this in mind, we have worked over the previous two weeks to begin a massive amount of data allocation. We began by creating an Excel sheet with every player from the 2012 NFL Draft, totaling over 300 players. Whenever we have free time available we have just been filling in these player's statistics from the NFL Combine and their college careers, so that within the next week we can use our massive data sheet to finally begin the actual work on our model. We are excited to have finally moved past a lot of the groundwork and clueless testing to finally be getting into the meat of the project.
|