Introduction

THE GREEN BAY PACKERS

Being from the great state of Wisconsin, one thing all Wisconsinites unite towards is to love and support the Green Bay Packers! I have gone to Packer games since I was a little kid, and I wanted to use this final project to look into the history of the Packers as well as to answers some questions I had about the 100 year history of the greatest football organization of all time. On top of this, I also wanted to analyze the recent NFL Draft and how fans viewed the Green Bay Packers picks. I thought this was really interesting because it allowed me to view not only the past Packers in this project, but also what people think about future Packers.For this research, I used Pro Football Reference and pulled data from the Packers past 100 years as a pro football team. Statistics included their record, postseason results, how many points they scored and allowed, and other analytics regarding the strength of their teams. For my secondary source, I used the Twitter API to scrape tweets about the Packers recent draft picks. I then used sentiment analysis to see what people thought of each pick.

The Data

Description of the Data used

Variable Description
Year Season Team Played
League What League the Team Played In
Team The star next to the team denotes that they made
the playoffs that year
Wins Regular Season Wins
Losses Regular Season Losses
Ties Regular Season Ties (They do happen)
Street_name Street name for the property
Div. Finish How the Team placed in their division
Playoffs How far the team went, if they made it
Points Scored Points Scored by Team for the Season
Points Allowed Points Allowed by the Team for the Season
Point Differential Points Scored - Points Allowed
Coaches Who was the coach that season
Approximate Value Which player statistically had the most value
to the team
Top Passer Best QB on Team
Top Rusher Top RB on Team
Top Receiver Top WR or TE on Team
Offensive Scoring Rank Where the Offense ranked in the NFL in scoring
Offensive Yardage Rank Where the Offense ranked in the NFL in yards
gained
Defensive Scoring Rank Where the Defense ranked in the NFL in scoring
allowed
Defensive Yards Rank Where the Defense ranked in the NFL in yards
allowed
Takeaway/Turnover Ratio Rank Where Packers ranked in NFL for their turnover
margin
Point Differential Rank Where Packers ranked in Point Differential
Yards Differential Rank Where the Packers ranked in yards gained-yards
allowed
Total Teams How many teams were in the NFL that year
Strength of Schedule Measures how good or difficult their opponents
were
Simple Rating System Margin of Victory - Strength of Schedule
Offensive SRS Packers offensive simple rating
Defensive SRS Packers defensive simple rating

The Table Itself

Average Amount of Wins Per Season

Average Amount of Wins per Season
7.742574

Franchise Winning Percentage

Franchise Winning Percentage
0.5581727

Average Offensive Scoring Rank

Average Offensive Scoring Rank
8.871287

Average Defensive Scoring Rank

Average Defensive Scoring Rank
9.712871

Most Dominant Season

Winningest Season

Worst Scoring Season

Least Successful Seasons

Descriptive Analysis

Question 1: How has the game of football changed for the Packers?

One thing we see in sports is how the games have changed over time. In basketball we have seen a three point shooting explosion, and in baseball we have seen differences in how long pitchers can pitch. In football, the main change has been from an old school running game to a more high-octane offensive football game. I wanted to see if this would show for the Packers, whether we would see a higher average of points scored and allowed per game in today’s age rather than in the early 1900s. To see this, I created two graphs that look at the average points per game and points allowed per game for every season. In the first graph, we see a positive relationship between year and average points per game. This means that as time has gone on and football has developed, scoring has increased on average for the Packers most years. A similar finding is shown in the second graph as average points allowed also increased over time. Another cool thing about these graphs is that we can pinpoint years or certain time periods where the Packers may have had a bad offense or a really poor defense. For example, in the 1950’s the Packers’ defense was the worst it has ever been, surrendering over 25 points per game. Since that stretch, the Packers defense has only allowed that many points per game in a season 3 times.

Question 2: Which Packers coach has had the most success?

In my time as a Packer fan, I have seen 4 different coaches lead the Green Bay Packers. However, I wanted to compare the success of some of these coaches to other past legendary Packer coaches to see which stand alone as the best in franchise history. To do this, I looked at total wins and total playoff appearances by coaches. After looking at these first two graphs, I took the top six coaches who had the most wins and playoff appearances: Matt LaFleur, Vince Lombardi, Curly Lambeau, Mike Holmgren, Mike Sherman and Mike McCarthy. I used these 6 coaches and made graphs depicting how many wins to losses they had per season as well as how they fared in the postseason. Looking at the graphs, Vince Lombardi and Curly Lambeau are far and away the best coaches in Packers history. However, I weas intrigued to see the success of Mike Holmgren, a coach who led the Packers in the 90’s and early 2000’s. Taking the Packers to two Super Bowls is not an easy feat, so his success definitely would make him one of the better coaches for the Packers in the modern era.

Question 3: What is more important for winning a chmapionship: offense or defense?

For this question, I wanted to look at whether having strong offensive play or defensive play was more important for the years that the Packers won championships. To do this, I filtered only the years the Packers won a championship and analyzed their average points per game and points allowed per game. On top of that I made a table looking at the average offensive and defensive ranks they had for each of these seasons. These ranks look at how the Packers ranked among the other teams in the NFL in Offensive scoring (Total Points Scored), offensive yardage(Total Yards Gained), defensive scoring (Total Points Allowed), and defensive yardage (Total Yards Allowed). The graph shows that many of the Packers championship teams were playing outstanding defense, only allowing at most 16 points per game. On the offensive side, most of the championship teams averaged at least 20 points a game. Looking at the table, most of the time the Packers title teams ranked better on defense than they did offense. While there are definitely more analytical statistics one could look at to determine factors for championship teams such as interceptions or sacks, this table gives a small glimpse at how being better at defense may be more important than offense for being a championship team.

Average Offensive Scoring Rank Average Offensive Yards Rank Average Defensive Scoring Rank Average Defensive Yards Rank
3.538461 4.769231 2.230769 2.846154

Question 4: Do the Analytical Statistics Tell a Good Story?

When I looked at this table, I saw a couple of statistics that I had never seen before. Called SRS, or Simple Rating System, the stat looks at the quality of a team compared to the average NFL team (0). Offensive and defensive SRS are also the Simple Rating System, but focus merely on their respective side of the football. MoV is Margin of Victory, and this is the average point differential per game. For example, if a team scored 300 points during this season and allowed 200, they would have a 100 point differential. Margin of Victory would take the point differential and divide it by games played. Lastly, strength of schedule looks at the average quality of opponents the team played by looking at the SRS system. Analyzing these stats, I wanted to see if any of them were a good indicator as to whether a team was actually good. I compared the offensive and defensive SRS ratings between Packer teams that made and did not make the playoffs and did the same thing with strength of schedule and margin of victory. Looking at the results, the two most important variables look to be offensive SRS and margin of victory as almost all the playoff teams had a positive number for these stats. The second graphs look at how the playoff teams fared in these statistics, wrapped by who far they progressed in the playoffs. From those results, the teams that ended up winning championships looked to have a much higher defensive SRS as well as a higher margin of victory, while the other two stats were relatively consistent with the other teams that lost in the postseason.

Question 5: Who is the Greatest Packer Quarterback Ever?

In the history of the Green Bay Packers, three quarterbacks have defined the organization with their accolades and dominance. These players are Bart Starr, Brett Favre, and Aaron Rodgers. I have had the privilege to watch Aaron Rodgers for all of my life, so I have always biasedly thought he was the best of the group. However, I want to look at accolades of these players such as how many times they have been to the playoffs, how the offense ranks on average with them at quarterback, and their success in the playoffs. The first graph looks at how many times each quarterback led their team to the playoffs. The results show that Aaron Rodgers and Brett Favre both hold the edge against Starr, going to the playoffs twice as many times as Starr had. The next graph though, looks at how each quarterback has fared in the postseason. The results here show that Starr has been the most successful, winning the most Super Bowls and championships while Favre and Rodgers have each only won one. The last table shows how the offense has ranked on average with each QB at the helm. This table shows that Starr and Favre had similar and better average rankings in terms of team offense in comparison to Rodgers. While one definitely would need to look more into the individual statistics to find a better glimpse at which quarterback was better, this data gives a good look at how each differenatiate themselves from one another in terms of achievments with the Pack.

Top Passer Average Offensive Scoring Rank Average Offensive Yards Rank
Favre 4.454546 6.909091
Rodgers 6.636364 9.454546
Starr 4.166667 6.833333

Secondary Data Source: Draft Night Extravaganza

Question 1: What words/topics were coming up the most in review of the Packers first round picks?

When the Packers selected their first two picks in the NFL draft, there werew a couple topics that came to mind that this table does reflect. The biggest talking point was that the Packers drafted two players from the University of Georgia, and this is seen as both tables have georgia as a recurring word. As you look more in the table, you see another Georgia player in Lewis Cine who was also taken in the first round. I found it interesting to see how many people were tweeting about the Georgia players as this became a story of the first round. However, another thing I found interesting about these tables was that more twitter users were comparing players at Quay Walker’s position than Devonte Wyatt. By this, I mean that many of people tweeted about players the Packers could have picked over Quay Walker such as Nakobe Dean and Devin Lloyd, both of whom play linebacker like Walker. In Devonte Wyatt’s case, not many users talked about other defensive lineman the Packers could have picked, showing how not many people had complaints of the pick at first glance. The only thing I was suprised about was not many tweets talking about a wide receiver.

Question # 2: How did Twitter view the Packers two picks from a sentiment analysis?

My second question looks at the sentiment behind the reaction to the two picks in the first round. I wanted to see what people thought about the two picks from an emotional standpoint, and for the most part many of the fans were happy. First, Devonte Wyatt’s sentiment analysis was very good to see as many fans “loved” the selection. Many talked about him being elite and a top talent. Of the negatives in his analysis, many of the words actually were pretty good. “Savage” is a pretty great term to refer to a football player as being tough and monstrous is what teams want. “Desperately” refers to the fact that the Packers really needed to fill up a whole at Wyatt’s position. The only two that could be somewhat negative are defensive and strange. These terms are used likely because many people thought the Packers were going to select a wide receiver, so many likely tweeted how they thought it was strange that the Packers picked another defensive player.

The second sentiment analysis looks at the reaction to the selection of Quay Walker. While love was used many times in this sentiment analysis, there also was much worse negative sentiment words used in reaction to this pick. This kind of reaction to this pick makes sense as many fans expected the Packers to take a wide receiver or even trade this pick in exchange for an already established player in the NFL. And with looking at the past table, we already knew that many Packers fans thought of other possible linebackers the Packers should have taken over Walker. It will be interesting to see how the sentiment analysis changes over the course of these two players first seasons.

How did the sentiment and topics change from Day 1 of the draft to Day 2?

Ok, so the issue I had with my time question was that all of my tweets came at around the same time. This made it really difficult to show real sentiment change over time. However, I was able to scrape data from Day 2 of the draft, in which the Packers turned the tide of sentiment. As seen in the table and word map, the Packers traded up in the second round with the Vikings and selected a wide receiver, Christian Watson. I was very intrigued to see how the fans reacted to this pick and to see as to whether or not the pick was as popular or negative as the Day 1 picks.

In the analysis, I found that people LOVED the Christian Watson pick. While there still was some negative sentiment towards the pick as some may have wanted another Georgia player (wide receiver George Pickens), the majority of sentiment showed fans being excited, happy, and loving the pick. I think this pick shows how fans can change opinions in the matter of one pick. Many fans left the first round skeptical of the Packers not drafting a wide receiver, but once they did it led to massive amounts of positive sentiment.

Regression Analysis: Picking What Matters in Winning a Title

For my predictive analysis, I wanted to see what metrics in my primary data set gave the best indication on what matters in winning a championship. Looking at my data, I took the primary metrics as seen below. My goal with the regression was to see what statistics were significant and non-significant until I found the statistics that were significant in predicting a championship winner. In order to do this, I made a column in the data, Championship, that measured to see if the Packers won a championship during a football season.

\[Championship = \alpha_i + Wins_i + Losses_i + Ties_i + \\ Points Scored_i +Points Allowed_i + Margin of Victory_i +\\ Strength of Schedule_i + Offensive SRS_i \\ \ + Defensive SRS_i + Takeaway/Turnover Ratio Rank_i\]

## 
## Call:
## lm(formula = Championship ~ Wins + Losses + Ties + `Margin of Victory` + 
##     `Strength of Schedule` + `Offensive SRS` + `Defensive SRS` + 
##     `Takeaway/Turnover Ratio Rank`, data = Packers)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -0.39623 -0.20572 -0.07306  0.10161  0.73987 
## 
## Coefficients:
##                                 Estimate Std. Error t value Pr(>|t|)
## (Intercept)                     0.155765   0.210370   0.740    0.461
## Wins                            0.001653   0.017327   0.095    0.924
## Losses                         -0.011426   0.022916  -0.499    0.619
## Ties                           -0.016639   0.052303  -0.318    0.751
## `Margin of Victory`            -0.003059   0.594592  -0.005    0.996
## `Strength of Schedule`          0.014706   0.595018   0.025    0.980
## `Offensive SRS`                 0.013691   0.593283   0.023    0.982
## `Defensive SRS`                 0.031584   0.594957   0.053    0.958
## `Takeaway/Turnover Ratio Rank` -0.001788   0.004823  -0.371    0.712
## 
## Residual standard error: 0.2949 on 92 degrees of freedom
## Multiple R-squared:  0.2938, Adjusted R-squared:  0.2324 
## F-statistic: 4.784 on 8 and 92 DF,  p-value: 6.15e-05

As seen above, my first regression was UGLY. None of the statistics were significant, so one by one I removed the statistics that were not significant in the data until I got to a final equation that had both variables be significant. I would have show regression by regression, but I removed all but two variables, so I thought I would save you all the time and just show you the final equation below.

\[Championship = \alpha_i + Offensive SRS_i + Defensive SRS_i\]

## 
## Call:
## lm(formula = Championship ~ `Offensive SRS` + `Defensive SRS`, 
##     data = Packers)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -0.37556 -0.20851 -0.07147  0.05904  0.78039 
## 
## Coefficients:
##                 Estimate Std. Error t value Pr(>|t|)    
## (Intercept)     0.065543   0.030730   2.133   0.0354 *  
## `Offensive SRS` 0.015264   0.006491   2.352   0.0207 *  
## `Defensive SRS` 0.035818   0.007659   4.677 9.34e-06 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.2878 on 98 degrees of freedom
## Multiple R-squared:  0.2832, Adjusted R-squared:  0.2685 
## F-statistic: 19.36 on 2 and 98 DF,  p-value: 8.235e-08

As seen in this regression, Offensive SRS and Defensive SRS were the most signficiant statistics when predicting a championship team. Even though all of the coefficients are low when measuring this, remember that Championship = 1 means the team would win a Super Bowl. I think what was especially fascinating about this was the concept of offense versus defense. In this regression model, it is shown that defensive statistics and defensive SRS are more important than offensive SRS and metrics, shown by the coefficient for defensive SRS being more than double of the Offensive SRS. This is an interesting finding and this preliminary analysis shows the value of defense and winning championships, which is why I personally believe the Packers drafted two defensive players in the first round instead of offense.

Conclusion

Overall, this project gave me a larger insight of the history of the Green Bay Packers, along with a better understanding of how teams find success in the NFL. For the Packers, their best seasons have been shown to involve having a strong defensive game. This correlates with their reasoning behind drafting defensive players early in the draft, something many people were upset about in the sentiment analysis. However, this project has allowed me to dive deep into the Packers history and see the players, coaches, and statistics that have made this organization the greatest football franchise in American football history.