NFL 2020 Preview with Python: Team Defense
NFL 2020 season is coming soon. For preview this season, I’m going to visualize some quarterbacks data using 2019 dataset.
1. Overview
In this article, I’m going to use this dataset as below. Thanks to Mr. Ron Yurko.
There is play-by-play dataset of pre-season, regular season and play-off. I’m going to use only regular season and visualize every team defensive stats, especially pass defense and run defense. This is brief visualization so please note that not kind of analysis and deeping dive.
Let’s get down to implementation.
About offense, I did about quarterback and rusher with individual stats. Please see also these article.
2. Preprocess
import pandas as pd
pd.set_option(“max_columns”, 400)
pbp = pd.read_csv(“play_by_play_data/regular_season/reg_pbp_2019.csv”)
games = pd.read_csv("games_data/regular_season/reg_games_2019.csv")Firstly, narrow down the columns which is needed for passing defense.
pbp_passing = pbp[
[
"game_id"
,"game_half"
,"qtr"
,"defteam"
,"down"
,"two_point_attempt"
,"yards_gained"
,"play_type"
,"first_down_pass"
,"pass_attempt"
,"complete_pass"
,"incomplete_pass"
,"sack"
,"touchdown"
,"interception"
,"pass_touchdown"
,"score_differential"
]
]And then, aggregate this dataset by defense team.
#Don't include two point attempt
pbp_passing.loc[pbp_passing.two_point_attempt == 1, "yards_gained"] = 0team_def_pass_stats = pbp_passing[
pbp_passing.pass_attempt == 1
].groupby(
"defteam"
,as_index=False
).agg(
{
"complete_pass": "sum"
,"yards_gained": "sum"
,"first_down_pass": "sum"
,"pass_touchdown": "sum"
,"incomplete_pass": "sum"
,"sack": "sum"
,"interception": "sum"
}
)team_def_pass_stats["pass_attempt"] = team_def_pass_stats["complete_pass"] + team_def_pass_stats["incomplete_pass"] + team_def_pass_stats["interception"]team_def_pass_stats["complete_rate"] = round(team_def_pass_stats["complete_pass"] / team_def_pass_stats["pass_attempt"], 3) * 100team_def_pass_stats = team_def_pass_stats_season[
[
"defteam"
,"pass_attempt"
,"complete_rate"
,"yards_gained"
,"pass_touchdown"
,"interception"
,"first_down_pass"
,"sack"
]
].sort_values("yards_gained").reset_index(drop=True)
On the other hand, I do the same for run defense.
Narrow down columns,
pbp_rushing = pbp[
[
"game_id"
,"game_half"
,"qtr"
,"down"
,"two_point_attempt"
,"defteam"
,"yards_gained"
,"play_type"
,"first_down_rush"
,"rush_attempt"
,"touchdown"
,"rush_touchdown"
,"score_differential"
]
]Aggregate by team.
team_def_rush_stats = pbp_rushing[
(pbp_rushing.rush_attempt == 1)
& (pbp_rushing.two_point_attempt == 0)
].groupby(
"defteam"
,as_index=False
).agg(
{
"rush_attempt": "sum"
,"yards_gained": "sum"
,"first_down_rush": "sum"
,"rush_touchdown": "sum"
}
).sort_values("yards_gained").reset_index(drop=True)
I want to use these dataset as one, so merge it. Merge key is “defteam”.
team_defense_stats = pd.merge(
team_def_pass_stats
,team_def_rush_stats
,on="defteam"
,how="inner"
,suffixes=["_passing", "_rushing"] #If there are same columns, add suffixes
)
In addition to this, I also need opponent points for every team. Extract this data from games dataset.
home_games = games.groupby("home_team", as_index=False).agg(
{"away_score": "sum"}
)home_games = home_games.rename(columns={"away_score": "opponent_points"})home_games = home_games.rename(columns={"home_team": "team"})away_games = games.groupby("away_team", as_index=False).agg({"home_score": "sum"})away_games = away_games.rename(columns={"home_score": "opponent_points"})away_games = away_games.rename(columns={"away_team": "team"})team_opponent_points = pd.merge(
home_games
,away_games
,on="team"
,how="inner"
,suffixes=["_home", "_away"]
)team_opponent_points["opponent_points"] = team_opponent_points.opponent_points_home + team_opponent_points.opponent_points_away
In the end, merge defense stats and opponent points dataset.
team_defense_stats = pd.merge(
team_defense_stats
,team_opponent_points
,left_on="defteam"
,right_on="team"
,how="inner"
)
3. Visualization
Let’s plot passing yards and rushing yards as scatter and color each plot depending on opponent points.
%matplotlib inline
import matplotlib.pyplot as pltwith
plt.rc_context(
{
"axes.edgecolor":"white"
,"xtick.color":"white"
, "ytick.color":"white"
, "figure.facecolor":"white"
}
):
fig = plt.figure(figsize=(30, 24), facecolor="black")
ax = fig.add_subplot(111, facecolor="black")#Plot scatter
s = ax.scatter(
team_defense_stats.yards_gained_passing
,team_defense_stats.yards_gained_rushing
,s=600
,alpha=0.5
,c=team_defense_stats.opponent_points
,cmap="bwr"
,marker="D"
)#Adjust looking
ax.set_xlabel("Pass Defense (Yds)", color="white", size=24)
ax.set_ylabel("Run Defense (Yds)", color="white", size=24)ax.set_xlim(5000, 2500) #less is better
ax.set_ylim(2500, 1000) #less is betterax.tick_params(axis="x", labelsize=24)
ax.tick_params(axis="y", labelsize=24)#Plot team name
for _, team in team_defense_stats.iterrows():
ax.text(
team.yards_gained_passing
,team.yards_gained_rushing
,team.defteam
,verticalalignment="center"
,horizontalalignment="center"
,fontsize=25
,color="white"
)#Colorbar settings
cb = plt.colorbar(s)
cb.set_label("Opponent Points (Season)", color="white", size=30)
cb.outline.set_edgecolor("white")
plt.setp(plt.getp(cb.ax.axes, 'yticklabels'), color="white")plt.title("Team Defensive Stats", color="white", size=30)
Patriots as “NE” is very good at both pass defense and run defense. Bills and Ravens also looks good. 49ers is strong at pass defense. They have less opponent points (blue marker) so we can say they are good as defense!
On the other hand, Buccaneers is outstanding at run defense but has more opponent points (red marker). Can we say that run defense less valuable than pass defense?
How is the breakdown of touchdown? Use pie chart.
plt.rc_context(
{
"axes.edgecolor":"white"
,"xtick.color":"white"
, "ytick.color":"white"
, "figure.facecolor":"white"
}
):
fig = plt.figure(figsize=(10, 10), facecolor="black")
ax = fig.add_subplot(111, facecolor="black")wedges, _, _ = ax.pie(
[pbp.pass_touchdown.sum(), pbp.rush_touchdown.sum()]
,labels=["Pass", "Rush"]
,textprops={
"color": "white"
,"fontsize": 20
}
,wedgeprops={"linewidth": 3}
,startangle=90
,counterclock=False
,autopct="%1.1f%%"
)ax.text(
0, 0
,str(int(pbp.pass_touchdown.sum() + pbp.rush_touchdown.sum()))
,color="white"
,ha="center"
,va="center"
,fontsize=40
)ax.set_title("Touchdown Ratio", color="white", size=40)plt.setp(wedges, width=0.2)
Could this mean preventing pass touchdown is more valuable than rush touchdown? I don’t think so. Of course passing can gain much more than rushing, so when it comes to scoring or touchdown, passing is likely chosen in the touchdown situation. Therefore, rushing looks less effective to get point numerally. It’s just common sense.
Take a look about which option, pass or run, is chosen in the touchdown situation.
Oh, I said “passing is likely chosen in the touchdown situation” but I wrong. Pass and run is equally chosen. Is Run defense less valuable than pass defense for preventing from opponent points, maybe this is wrong issue and wasting time to answer this. We need deeper dive about defense.
By the way, I know you think that how can we plot team logo instead of marker? We can do that using “artist”. To get more details, please check my article.
Of course, you need to prepare all of team image file in previous. This is much harder than programming ;)
from matplotlib.offsetbox import OffsetImage, AnnotationBboxplt.rc_context(
{
"axes.edgecolor":"white"
,"xtick.color":"white"
, "ytick.color":"white"
, "figure.facecolor":"white"
}
):
fig = plt.figure(figsize=(30, 24), facecolor="black")
ax = fig.add_subplot(111, facecolor="black")for _, team in team_defense_stats.iterrows():
#read image file
image = plt.imread(root + "image/" + str(team.defteam) + ".png")
ax.add_artist( #ax can be added image as artist.
AnnotationBbox(
OffsetImage(image)
,(team.yards_gained_passing, team.yards_gained_rushing)
,frameon=False
)
)ax.set_xlabel("Pass Defense (Yds)", color="white", size=24)
ax.set_ylabel("Run Defense (Yds)", color="white", size=24)ax.set_xlim(5000, 2500) #less is better
ax.set_ylim(2500, 1000) #less is betterax.tick_params(axis="x", labelsize=24)
ax.tick_params(axis="y", labelsize=24)plt.title("Team Defensive Stats", color="white", size=30)
Looks better? It’s easy to understand where team is but I think marker with color bar is better than team logo in terms of data visualization including more information.
Thank you for reading.
