Total Attendance Breakdown

As mentioned in the introduction, this data analysis will look into if the number of fans in attendance correlates to a team’s success (number of games won). Additionally, it will provide comparisons of how teams fare in home versus away games while keeping in mind the attendance at those games. Earlier in the data preparation, I split the attendance dataset into two separate datasets, nfl_total_attendance and nfl_weekly_attendance. To first understand the importance of fan attendance, it is critical to observe which teams have the strongest fan base over the past 20 years.

Instead of using the teams’ total attendance numbers, I wanted to take an average of each team’s weekly attendance. I feel this will give me a more accurate representation of attendance. With that said, I added a column to the nfl_weekly_attendance column to calculate the mean.

Using the ggplotly function, the graph becomes interactive. To look and interact with the visualization to the right, you can scroll over the lines to get a detailed description including the year, total attendance, and team. You can click on a team once to remove it from the visualization, or you can double-click on the team in the legend to isolate that line. This interaction enables you to filter to specific teams in order to see their attendance trends since 2000.

In the visualization to the right, it is evident that the Dallas Cowboys appear to have the strongest fan base, and the Los Angeles Chargers appear to have the weakest fan base. The top five teams with the current highest attendance records are:

1. Dallas Cowboys

2. Green Bay Packers

3. Los Angeles Rams

4. New York Giants

5. Philadelphia Eagles

It is also important to note that the spike in attendance for the Dallas Cowboys in 2009 can be attributed to the opening of their brand new AT&T Stadium. This stadium opened on May 27, 2009. The stadium holds 80,000 people in the stands but can be expanded to hold more than 100,000 individuals when standing room only areas are included.

Offense vs. Defense

In the next step of this analysis, I investigated the importance of an offense and a defense. The offense in football is the 11 players who are on the field for a team when they have the ball. Conversely, the defense is the 11 players on the field when the other team has the ball. Sports writers and analysts have argued over the years whether a better offense or defense is more critical to a team’s success. Utilizing the nfl_standings dataset, I sought to analyze this discussion.

First, I created a linear model showcasing a team’s wins in a season using offensive_ranking and defensive_ranking as the predictor variables.

# Attach the dataset
attach(nfl_standings)

# Create a linear model
rankings_model <- lm(total_wins ~ offensive_ranking + defensive_ranking)
summary(rankings_model)

Call:
lm(formula = total_wins ~ offensive_ranking + defensive_ranking)

Residuals:
    Min      1Q  Median      3Q     Max 
-4.2881 -1.0127 -0.0189  1.0536  5.0782 

Coefficients:
                  Estimate Std. Error t value Pr(>|t|)    
(Intercept)        7.98430    0.05741  139.06   <2e-16 ***
offensive_ranking  0.44409    0.01363   32.58   <2e-16 ***
defensive_ranking  0.43527    0.01657   26.26   <2e-16 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Residual standard error: 1.486 on 667 degrees of freedom
Multiple R-squared:  0.7711,    Adjusted R-squared:  0.7705 
F-statistic:  1124 on 2 and 667 DF,  p-value: < 2.2e-16

The model showcases that both offensive_ranking and defensive_ranking are significant variables in determining a team’s total wins at a 99% confidence level. To drill deeper, the correlation coefficients were discovered for each predictor variable to total wins.

# Run correlation tests
cor(offensive_ranking, total_wins)

[1] 0.7310874

cor(defensive_ranking, total_wins)

[1] 0.6379049

The offensive_ranking had a coefficient of 0.7311 and the defensive_ranking had a coefficient 0.6379. As such, it appears that a team’s offense has a greater correlation to a team’s wins than its defense. To visualize this, I plotted two graphs to further test this hypothesis. These can be seen in the tabs to the right.

These graphs confirm the positive correlation between an increasing offensive or defensive ranking and a team’s win. Additionally, the confidence band in the defensive ranking is larger than the offensive ranking’s band. This agrees with my conclusion that the offensive’s ranking correlation is stronger than the defense.

Points For vs. Points Against

Using different statistics now, I changed the predictor variables to be points_for and points_against as these represent offensive and defensive success, respectively. Then, I used the binary playoffs variable to see how scoring or giving up points led to a team’s probability of making the playoffs. I took the same approach as the previous variables.

This model shows that points_for and points_against are both significant as well to a team’s total wins. Additionally, the correlations to total wins are 0.7276 and -0.6667. This indicates a strong, positive relationship for points_for and a strong, negative relationship for points_against. The offensive side, once again, has a slightly stronger relationship.

The graphical representations show these strong linear relationships as well with playoff teams typically having a low Total Points Against and high Total Points For on the season.

# Create linear model
points_model <- lm(total_wins ~ points_for + points_against)
summary(points_model)

Call:
lm(formula = total_wins ~ points_for + points_against)

Residuals:
    Min      1Q  Median      3Q     Max 
-3.3942 -0.8523 -0.0140  0.8298  3.8933 

Coefficients:
                 Estimate Std. Error t value Pr(>|t|)    
(Intercept)     8.5301159  0.4048277   21.07   <2e-16 ***
points_for      0.0274469  0.0006803   40.34   <2e-16 ***
points_against -0.0289973  0.0008114  -35.74   <2e-16 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Residual standard error: 1.248 on 667 degrees of freedom
Multiple R-squared:  0.8385,    Adjusted R-squared:  0.838 
F-statistic:  1732 on 2 and 667 DF,  p-value: < 2.2e-16

# Run correlation tests
cor(points_for, total_wins)

[1] 0.7275527

cor(points_against, total_wins)

[1] -0.6666812

Can Weather Predict Game Outcomes?

Next, I wanted to see in high wind speeds were correlated to low-scoring games. To do this, I first combined the total score of the home_score and away_score variables. I created a total_score column. With this column, I ran correlation coefficients between total_score and temperature, total_score and humidity, and total_score and wind_mph.

I also ran two linear models to see if weather conditions could predict whether or not the game would be high-scoring or low-scoring. I was able to train and test my dataset with a 70-30 training-testing split.

It appears that the correlation coefficient between total_score and temperature is 0.0164. Knowing that a correlation coefficient value of plus or minus one is said to be a perfect correlation, I know that there is little to no correlation between total_score and temperature.

The correlation coefficient between total_score and humidity is -0.1207. Here, I can see that there is a slight, negative correlation between the two variables. Similar to this is the relationship between total_score and wind_mph. The correlation coefficient is -0.1328, indicating a slight, negative correlation.

Of the three relationships tested, it appears that the wind speed correlates most to a lower total score. The higher the wind, the lower the combined score of the game.

Looking further, I decided to create a linear model to see if temperature, humidity, and wind can predict the total combined score of the game. The first model I created included all three predictor variables; however, it did not perform well. The adjusted R-squared of this model is around a meager 0.02, indicating that the model accounts for only 2% of the variance explained by the model. It did appear that the humidity and wind_mph variables were statistically significant at the 95% confidence level as they had p-values less than 0.05. The Mean Squared Error (MSE) of this model is higher than 170, which is incredibly high given one would ideally want an MSE of zero. It is important to note that these results will vary given the random training-testing split.

Using the above data and knowing that wind_mph was the most correlated with total_score, I decided to create a second model with just wind_mph as the sole predictor variable. This model performed even worse with an adjusted R-squared of around 0.015, indicating that the model accounts for less than 2% of the variance. The wind_mph variable is still statistically significant due to its p-value being less than 0.05 at the 95% confidence interval. Additionally, the MSE of this model is still higher than 170. It is important to note that these results will also vary given the random training-testing split.

In conclusion, weather does not accurately predict whether or not the game will be high-scoring or low-scoring. I originally thought it would be more difficult for the players to score given higher wind speeds; however, I was proved wrong.

I thought about analyzing how teams fared in games located on opposite sides of the country (e.g., New England Patriots at Los Angeles Rams); however, I decided against that analysis. I decided against this because I thought there would be other contributing factors to a loss, such as home-field advantage.

Successful Postseason Teams

For the next part of my analysis, I decided to look at the best head coaches and quarterbacks from the past 20 years. I created this nfl_playoffs dataset myself by taking every team from the past 20 years that made the playoffs and then listing their head coach and starting postseason quarterback.

The goal of this portion of the analysis is to see if the head coach or quarterback really do make a difference in team success. For example, was it a coincidence that the Tampa Bay Buccaneers won the most recent Super Bowl – conveniently, the first year with Tom Brady as quarterback?

Over the past 20 years, it appears that the top three teams are as follows:

1. New England Patriots

2. Indianapolis Colts

3. Green Bay Packers

The New England Patriots have been a dominant force the past few decades. I would argue that everyone that is not a New England Patriots fan strongly roots against them just because they have won so frequently. In the past few years, fans would claim the success comes from head coach Bill Belichick and former players – Tom Brady and Rob Gronkowski.

The Indianapolis Colts’ success over the past 20 years can be attributed to their former quarterback Peyton Manning. Manning arrived to the Colts in 1998 and led the team to its first championship in 36 seasons at Super Bowl XLI.

The Green Bay Packers have also been quite the team over the past 20 years. Their general manager, Ted Thompson, has been a key figure in their success. The Packers also have had incredible leaders through their coaching staff and players. Notably, the Packers also have a notoriously strong fan base.

Successful Postseason Head Coaches

Now knowing the top-performing teams in the NFL, I want to see which coaches have led these teams to success.

Looking at the visualization Top Ten Coaches by Playoff Appearances to the right, it is evident that the top two coaches in the NFL over the past 20 years are:

1. Bill Belichick

2. Andy Reid

Bill Belichick, head coach of the New England Patriots, has been with the team since 2000. As head coach, he has six Super Bowl championships (XXXVI, XXXVIII, XXXIX, XLIX, LI, and LIII). He has won AP NFL Coach of the Year in 2003, 2007, and 2010. He has also won 31 playoff games. I cannot say I was surprised to see him listed as number one in this analysis. His career record as a coach is 311-148 (0.678).

Andy Reid is the current head coach for the Kansas City Chiefs. He has been with the team since 2013. Prior to that, Reid was the head coach for the Philadelphia Eagles (1999 - 2012). He has won two Super Bowl championships (XXXI and LIV) - one as an assistant coach and one as a head coach. His career record as a coach is 238-145-1 (0.621).

Next, I wanted to look at the head coaches with the most Super Bowl championships won. Is this consistent with the top coaches who go to postseason play? Given Marvin Lewis, former head coach of the Cincinnati Bengals, is number ten in the graphic Top Ten Coaches by Playoff Appearances, I cannot be quite sure.

In the visualization Top Ten Coaches by Super Bowl Championships to the right, it is evident that Bill Belichick is still the dominant head coach from the past 20 years, with six Super Bowl championships.

The head coach with the next highest number of Super Bowl championships as head coach is Tom Coughlin. Tom Coughlin was the head coach of the Jacksonville Jaguars from 1995 - 2002, and he was the head coach of the New York Giants from 2004 - 2015. His two Super Bowl championships were as the head coach of the New York Giants (XLII and XLVI). His career record as a coach in the NFL was 182-157 (0.537).

Successful Postseason Quarterbacks

An individual can be a great coach; however, they also need a great team. Quarterbacks are often described as the leaders of the NFL. With that said, I wanted to take a look at the best quarterbacks from the past 20 years.

From the visualization Top Ten Quarterbacks by Playoff Appearances to the right, it is evident that the top three quarterbacks in the NFL over the past 20 years are:

1. Tom Brady

2. Ben Roethlisberger

3. Drew Brees

It is no surprise that Tom Brady is number one, as he is often referred to as the Greatest Of All Time (GOAT). Tom Brady was drafted to the New England Patriots in the sixth round of the 2000 NFL Draft. Since then, he is a seven-time Super Bowl Champion (six with the New England Patriots and one with the Tampa Bay Buccaneers). He is still active in the NFL as the quarterback for the Tampa Bay Buccaneers. As of 2020, his completion percentage is 64% and his accolades are many. I am interested to see how much longer he will excel in the league.

Ben Roethlisberger, the long-time quarterback for the Pittsburgh Steelers, was drafted in the first round of the 2004 NFL Draft. He has won two Super Bowl championships, and his completion percentage is 64.4%. He just signed with the Steelers for another year, so (as a Steelers fan) I am hoping he will lead the team to a third championship this upcoming season.

Drew Brees started his career with the San Diego Chargers (2001 - 2005), but he is most known for his career as the quarterback for the New Orleans Saints (2006 - 2020). Brees has won one Super Bowl championship, and he just announced his retirement this year. His completion percentage was 67.7%.

These top three quarterbacks are some of the famous players in the NFL. It is interesting to see how their career statistics speak for themselves.

Next, I wanted to see which quarterbacks have won the most Super Bowl Championships over the past 20 years.

Once again, looking at the Top Ten Quarterbacks by Super Bowl Championships to the right, Tom Brady is the most dominant quarterback in the NFL based on Super Bowl Championships. Ben Roethlisberger is close behind. The other two quarterbacks which I have not discussed are Eli Manning and Peyton Manning. Both of them have won two Super Bowl Championships since 2000.

Eli Manning was the quarterback for the New York Giants from 2004 - 2019. He won two Super Bowl Championships (XLII and XLVI) and was the Super Bowl MVP for both games.

Peyton Manning was the quarterback for the Indianapolis Colts from 1998 - 2011 and the quarterback for the Denver Broncos from 2012 - 2015. He also won two Super Bowl Championships (XLI, 50) and was the Super Bowl MVP for Super Bowl XLI. He won one Super Bowl with the Colts in 2006 and one Super Bowl withe the Broncos in 2015.

Passing Yards Leaders

For the next portion of my analysis, I wanted to analyze the top passing yards leaders. Given these are always quarterbacks, I wanted to see if this was consistent with my previous analysis of postseason quarterback success.

From the visualizations to the right, it is evident that Drew Brees was the leader for passing yards six out of 20 times the past 20 years. However, he did not have the most playoff appearances. It is interesting that Tom Brady is on the list only three times, yet he has by far been the most successful quarterback.

Additionally, looking at the Top Passing Yards Leaders tab, the visualization is interactive. It is evident that the top passer over the past 20 years was Peyton Manning in 2013. The data points are colored based on player. The legend can be seen on the right.

Rushing Yards Leaders

Looking at the leaders for rushing yards, I performed the same analysis as above.

As seen in the visualizations to the right, the rushing yards leaders have much more variation than the passing yards leaders. Most recently, Derrick Henry from the Tennessee Titans has been the dominant running back with 2027 yards in 2020. He has been the top rusher for the past two years in a row.

Additionally, looking at the Top Rushing Yards Leaders tab, the visualization is also interactive. It is evident that the top rusher over the past 20 years was Adrian Peterson in 2012. The data points are colored based on player. The legend can be seen on the right.

Rival Analysis

The last part of my analysis is to look at the top five rivals in the NFL and the performance of these games over the past 20 years.

The term “rivalry” can be a bit subjective; however I chose the following five rivalries to analyze:

1. Green Bay Packers vs. Chicago Bears – It is evident that the Green Bay Packers have been dominant in this rivalry, winning 71% of the encounters.

2. Dallas Cowboys vs. Philadelphia Eagles – This is a very good rivalry, as both teams have performed. It is evident that the Philadelphia Eagles have been leading this rivalry, winning 54% of the encounters.

3. Kansas City Chiefs vs. Las Vegas Raiders (formerly Oakland Raiders) – It is evident that the Kansas City Chiefs have been dominant in this rivalry, winning 63% of the encounters.

4. Baltimore Ravens vs. Pittsburgh Steelers – This is a phenomenal rivalry to watch, as both times have won 22 games out of 44 games total. Each team has won 50% of the encounters.

5. Washington Football Team (formerly Washington Redskins) vs. New York Giants – It is evident that the New York Giants have been dominant in this rivalry, winning 68% of the encounters.

Then, just for fun – I decided to analyze the Pittsburgh Steelers vs. the Cincinnati Bengals for fun. It is evident that the Pittsburgh Steelers have been dominant in this rivalry, winning 79% of the encounters.