How they actually go:¶
1) Figure out how you're going to get data.¶
2) Spend a lot of time cleaning data.¶
3) Look at what are supposed to be numbers.¶
4) Spend more time cleaning and getting data that you found out you need.¶
5) Visualize.¶
6) Clean + Splice creatively.¶
7) DO MATH!¶
8) Rinse and repeat steps 3-7 until deadline,¶
For everyone's sake, we're not really going to get into steps 1 and 2.
Step 1: Get the data.¶
Forget what I said earlier - this is the hardest part.¶
Thankfully, smart people are working on problems like this, so for the purposes of this talk, it's going to be solved by nba_py: https://github.com/seemethere/nba_py¶
In [4]:
import nba_helper_functions
from nba_py import team, game
import pandas as pd
import numpy as np
import plotly.plotly as py
import plotly.graph_objs as go
import matplotlib.pyplot as plt
import seaborn as sns
import pandas as pd
import nbashots as nba # this will throw a warning if using matplotlib 1.5
import requests as r
%matplotlib inline
In [5]:
cavs_id = 1610612739
Step 2: Splice and clean the data:¶
Writing helper functions counts as cleaning.
In [6]:
def lebron_data(season, season_type):
"""Gets info on all cavs games, and returns dataframe with only Lebron's statistics"""
val = team.TeamGameLogs(cavs_id, season=season, season_type = season_type)
game_ids = val.info()['Game_ID'].tolist()
data = pd.DataFrame()
for id in game_ids:
all_players = game.PlayerTracking(id).info()[['GAME_ID', 'PLAYER_ID', 'MIN', 'SPD', 'TCHS', 'PASS', 'AST', 'CFGM', 'CFGA', 'UFGM', 'UFGA']]
lebron = all_players.loc[all_players['PLAYER_ID'] == 2544]
#this gets some player tracking stats; average speed, touches, passes, assists, and some shot info.
data = data.append(lebron)
return data
In [7]:
df_playoffs = pd.DataFrame()
playoff_seasons = ['2014-15', '2015-16', '2016-17']
for season in playoff_seasons:
df = lebron_data(season, 'Playoffs')
df_playoffs = df_playoffs.append(df)
In [8]:
df_regular_season = lebron_data('2016-17', 'Regular Season')
In [9]:
df_playoffs.head()
Out[9]:
Now that we've gotten a good chunk, it's time to clean!
In [11]:
#Filtering out rest games
df_regular_season = df_regular_season[df_regular_season['SPD'] != 0]
In [16]:
def hh_mm_ss2seconds(hh_mm_ss):
import functools
return functools.reduce(lambda min, sec: (min*60 + sec)/60.0, map(int, hh_mm_ss.split(':')))
In [17]:
df_playoffs['MIN'] = df_playoffs['MIN'].map(lambda x: hh_mm_ss2seconds(x))
df_regular_season['MIN'] = df_regular_season['MIN'].map(lambda x: hh_mm_ss2seconds(x))
Step 3: Numbers!¶
In [18]:
df_playoffs.describe()
Out[18]:
In [19]:
df_regular_season.describe()
Out[19]:
Step 4: Some more splicing:¶
In [20]:
df_playoffs.describe() - df_regular_season.describe()
Out[20]:
(We're skipping step 5 for now- it's not fun.)¶
Step 6: Time to clean+visualize (again):¶
In [21]:
court_shapes = nba_helper_functions.court_shapes()
In [23]:
playoff_df = pd.DataFrame()
for season in playoff_seasons:
df = nba_helper_functions.get_shot_info(season= season)
playoff_df = playoff_df.append(df)
In [25]:
missed_shot_trace = go.Scatter(
x = playoff_df[playoff_df['EVENT_TYPE'] == 'Missed Shot']['LOC_X'],
y = playoff_df[playoff_df['EVENT_TYPE'] == 'Missed Shot']['LOC_Y'],
mode = 'markers',
name = 'Missed Shot',
marker = dict(
size = 5,
color = 'rgba(255, 255, 0, .8)',
line = dict(
width = 1,
color = 'rgb(0, 0, 0, 1)'
)
)
)
made_shot_trace = go.Scatter(
x = playoff_df[playoff_df['EVENT_TYPE'] == 'Made Shot']['LOC_X'],
y = playoff_df[playoff_df['EVENT_TYPE'] == 'Made Shot']['LOC_Y'],
mode = 'markers',
name = 'Made Shot',
marker = dict(
size = 5,
color = 'rgba(0, 200, 100, .8)',
line = dict(
width = 1,
color = 'rgb(0, 0, 0, 1)'
)
)
)
layout = go.Layout(
title='Shots by Lebron Last 3 Playoff Series',
showlegend=True,
xaxis=dict(
showgrid=False,
range=[-300, 300]
),
yaxis=dict(
showgrid=False,
range=[-100, 500]
),
height=600,
width=650,
shapes=court_shapes
)
data = [missed_shot_trace, made_shot_trace]
fig = go.Figure(data=data, layout=layout)
py.iplot(fig, filename='Lebron Playoffs Shot Chart')
Out[25]:
In [26]:
regular_season = nba_helper_functions.get_shot_info(season_type = 'Regular Season')
In [28]:
missed_shot_trace = go.Scatter(
x = regular_season[regular_season['EVENT_TYPE'] == 'Missed Shot']['LOC_X'],
y = regular_season[regular_season['EVENT_TYPE'] == 'Missed Shot']['LOC_Y'],
mode = 'markers',
name = 'Missed Shot',
marker = dict(
size = 5,
color = 'rgba(255, 255, 0, .8)',
line = dict(
width = 1,
color = 'rgb(0, 0, 0, 1)'
)
)
)
made_shot_trace = go.Scatter(
x = regular_season[regular_season['EVENT_TYPE'] == 'Made Shot']['LOC_X'],
y = regular_season[regular_season['EVENT_TYPE'] == 'Made Shot']['LOC_Y'],
mode = 'markers',
name = 'Made Shot',
marker = dict(
size = 5,
color = 'rgba(0, 200, 100, .8)',
line = dict(
width = 1,
color = 'rgb(0, 0, 0, 1)'
)
)
)
layout = go.Layout(
title='Shots by Lebron 2016-2017 Regular Season',
showlegend=True,
xaxis=dict(
showgrid=False,
range=[-300, 300]
),
yaxis=dict(
showgrid=False,
range=[-100, 500]
),
height=600,
width=650,
shapes=court_shapes
)
data = [missed_shot_trace, made_shot_trace]
fig = go.Figure(data=data, layout=layout)
py.iplot(fig, filename='Lebron Regular Season Shot Chart')
Out[28]:
In [29]:
from scipy import stats
In [30]:
mean, xedges, yedges, binnumber = stats.binned_statistic_2d(x=playoff_df.LOC_X, y=playoff_df.LOC_Y, values=playoff_df.SHOT_MADE_FLAG, statistic='mean', bins=35)
In [31]:
data = [
go.Heatmap(
z=mean.T,
x=xedges,
y=yedges,
colorscale='YIOrRd',
opacity = .5,
reversescale = True
)
]
layout = go.Layout(
title='Playoff Lebron Binned Heatmap Last 3 Years',
shapes=court_shapes
)
fig = go.Figure(data=data, layout=layout)
py.iplot(fig, filename='Playoff Lebron Binned Heatmap Last 3 Years')
Out[31]:
In [32]:
mean_reg, xedges_reg, yedges_reg, binnumber = stats.binned_statistic_2d(x=regular_season.LOC_X, y=regular_season.LOC_Y, values=regular_season.SHOT_MADE_FLAG, statistic='mean', bins=35)
data = [
go.Heatmap(
z=mean_reg.T,
x=xedges_reg,
y=yedges_reg,
colorscale='YIOrRd',
opacity = .5,
reversescale = True
)
]
layout = go.Layout(
title='Lebron Binned Heatmap 2016-17',
shapes=court_shapes
)
fig = go.Figure(data=data, layout=layout)
py.iplot(fig, filename='Lebron Binned Heatmap 2016-17')
Out[32]:
In [45]:
from matplotlib.offsetbox import OffsetImage
grid = nba.shot_chart_jointplot(playoff_df.LOC_X, playoff_df.LOC_Y,
title="Lebron Last 3 Playoffs")
plt.show()
In [46]:
grid = nba.shot_chart_jointgrid(regular_season.LOC_X, regular_season.LOC_Y,
title="Lebron 2016-2017 Regular Season")
plt.show()
In [52]:
cmap=plt.cm.gist_heat_r
grid = nba.shot_chart_jointgrid(regular_season.LOC_X, regular_season.LOC_Y,
title="Lebron 2016-2017 Regular Season",marginals_color=cmap(.7),
joint_type="scatter",joint_kde_shade=True)
grid.plot_joint(sns.kdeplot, cmap=plt.cm.OrRd_r)
plt.show()
In [54]:
cmap=plt.cm.gist_heat_r
grid = nba.shot_chart_jointgrid(playoff_df.LOC_X, playoff_df.LOC_Y,
title="Lebron Last 3 Playoffs",
joint_type="scatter", marginals_color=cmap(.3), joint_kde_shade=True)
grid.plot_joint(sns.kdeplot, cmap=plt.cm.OrRd_r)
plt.show()