In [108]:

#Import the libraries and packages we need
import numpy as np
from sklearn.linear_model import LinearRegression
import matplotlib.pyplot as plt
import pandas as pd
import highlight_text

In [103]:

# the famous import font code to use Andale Mono
import matplotlib.font_manager
from IPython.core.display import HTML

def make_html(fontname):
    return "<p>{font}: <span style='font-family:{font}; font-size: 24px;'>{font}</p>".format(font=fontname)

code = "\n".join([make_html(font) for font in sorted(set([f.name for f in matplotlib.font_manager.fontManager.ttflist]))])

In [56]:

df = pd.read_csv('1995_96 - 2020 Premier league standings.csv')

In [57]:

df

Out[57]:

	Rk	Squad	MP	W	D	L	GF	GA	Pts	Attendance	Top Team Scorer	Goalkeeper	Year
0	1	Manchester Utd	38	25	7	6	73	35	82	41001	Eric Cantona - 14	Peter Schmeichel	1996
1	2	Newcastle Utd	38	24	6	8	66	37	78	36501	Les Ferdinand - 25	Shaka Hislop	1996
2	3	Liverpool	38	20	11	7	70	34	71	39553	Robbie Fowler - 28	David James	1996
3	4	Aston Villa	38	18	9	11	52	35	63	37492	Dwight Yorke - 17	Mark Bosnich	1996
4	5	Arsenal	38	17	12	9	49	32	63	32614	Ian Wright - 15	David Seaman	1996
...	...	...	...	...	...	...	...	...	...	...	...	...	...
495	16	West Ham	38	10	9	19	49	62	39	44155	Michail Antonio - 10	Łukasz Fabiański	2020
496	17	Aston Villa	38	9	8	21	41	67	35	28505	Jack Grealish - 8	Tom Heaton	2020
497	18	Bournemouth	38	9	7	22	40	65	34	7745	Callum Wilson - 8	Aaron Ramsdale	2020
498	19	Watford	38	8	10	20	36	64	34	15353	Troy Deeney - 10	Ben Foster	2020
499	20	Norwich City	38	5	6	27	26	75	21	19913	Teemu Pukki - 11	Tim Krul	2020

500 rows × 13 columns

In [58]:

df['GD'] = df['GF'] - df['GA']

In [59]:

df

Out[59]:

	Rk	Squad	MP	W	D	L	GF	GA	Pts	Attendance	Top Team Scorer	Goalkeeper	Year	GD
0	1	Manchester Utd	38	25	7	6	73	35	82	41001	Eric Cantona - 14	Peter Schmeichel	1996	38
1	2	Newcastle Utd	38	24	6	8	66	37	78	36501	Les Ferdinand - 25	Shaka Hislop	1996	29
2	3	Liverpool	38	20	11	7	70	34	71	39553	Robbie Fowler - 28	David James	1996	36
3	4	Aston Villa	38	18	9	11	52	35	63	37492	Dwight Yorke - 17	Mark Bosnich	1996	17
4	5	Arsenal	38	17	12	9	49	32	63	32614	Ian Wright - 15	David Seaman	1996	17
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
495	16	West Ham	38	10	9	19	49	62	39	44155	Michail Antonio - 10	Łukasz Fabiański	2020	-13
496	17	Aston Villa	38	9	8	21	41	67	35	28505	Jack Grealish - 8	Tom Heaton	2020	-26
497	18	Bournemouth	38	9	7	22	40	65	34	7745	Callum Wilson - 8	Aaron Ramsdale	2020	-25
498	19	Watford	38	8	10	20	36	64	34	15353	Troy Deeney - 10	Ben Foster	2020	-28
499	20	Norwich City	38	5	6	27	26	75	21	19913	Teemu Pukki - 11	Tim Krul	2020	-49

500 rows × 14 columns

In [60]:

y=df.Rk
x=df.GD.values.reshape(-1,1)

In [61]:

x.shape

Out[61]:

(500, 1)

In [62]:

y.shape

Out[62]:

(500,)

In [63]:

model = LinearRegression().fit(x,y)

In [64]:

r_sq = model.score(x,y)
intercept = model.intercept_
slope = model.coef_

In [65]:

y_pred = intercept + slope*x

In [126]:

fig,ax = plt.subplots(figsize=(10,10))
fig.set_facecolor('#f3edd3')
ax.patch.set_facecolor('#f3edd3')

ax.grid(ls='dotted',lw=.5,color='lightgrey',axis='y',zorder=1)
spines = ['top','right']
for i in spines:
    if i in spines:
        ax.spines[i].set_visible(False)

#plot data
plt.scatter(x,y,alpha=.7)
plt.plot(x,y_pred,c='red',linestyle='--',dashes=(5,5))


plt.ylim(.5,20.5)
plt.gca().invert_yaxis()

plt.yticks([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20])

plt.xlabel('Goal Differential',fontsize=18,fontname='Andale Mono')
plt.ylabel('Final Leage Rank',fontsize=18,fontname='Andale Mono')
plt.title('Evaluating the Relationship Between Goal Differential and League Table Rank\nPremier League Seasons 1995/96 - 2019/2020',ha='center',fontsize=16,fontname='Andale Mono')

#annotate teams
ax.annotate(xy=(79,.7),text='Man City,2017/18')
ax.annotate(xy=(-75,19.5),text='Derby County,2007/08')


#analysis
ax.annotate(xy=(-65,1.65),text=f'R-Squared = {round(r_sq,2)}\nThe regression equation: y = {intercept} + {slope} * x ',fontname='Andale Mono')
#ax.annotate(xy=(10,9),text='No team that finished 9th underperformed the model',fontname='Andale Mono')
#ax.annotate(xy=(-5,17),text='No team that finished 17th overperformed the model',fontname='Andale Mono')
ax.annotate(xy=(-65,3),text=f'The model fits the data very well.\nInterestingly enough, the spread of all the data points\nfor each rank is fairly small.',fontname='Andale Mono')

plt.savefig('premline.png',dpi=300,bbox_inches = 'tight',facecolor='#f3edd3')

In [129]:

df.sort_values(by='Rk',ascending=True).head(30)

Out[129]:

	Rk	Squad	MP	W	D	L	GF	GA	Pts	Attendance	Top Team Scorer	Goalkeeper	Year	GD
0	1	Manchester Utd	38	25	7	6	73	35	82	41001	Eric Cantona - 14	Peter Schmeichel	1996	38
440	1	Manchester City	38	32	4	2	106	27	100	54070	Sergio Agüero - 21	Ederson	2018	79
60	1	Manchester Utd	38	22	13	3	80	37	79	55188	Dwight Yorke - 18	Peter Schmeichel	1999	43
420	1	Chelsea	38	30	3	5	85	33	93	41508	Diego Costa - 20	Thibaut Courtois	2017	52
80	1	Manchester Utd	38	28	7	3	97	45	91	55107	Dwight Yorke - 20	Mark Bosnich	2000	52
400	1	Leicester City	38	23	12	3	68	36	81	31998	Jamie Vardy - 24	Kasper Schmeichel	2016	32
100	1	Manchester Utd	38	24	8	6	79	31	80	67489	Teddy Sheringham - 15	Fabien Barthez	2001	48
380	1	Chelsea	38	26	9	3	73	32	87	41546	Diego Costa - 20	Thibaut Courtois	2015	41
120	1	Arsenal	38	26	9	3	79	36	87	38055	Thierry Henry - 24	David Seaman	2002	43
360	1	Manchester City	38	27	5	6	102	37	86	47075	Yaya Touré - 20	Joe Hart	2014	65
140	1	Manchester Utd	38	25	8	5	74	34	83	67602	Ruud van Nistelrooy - 25	Fabien Barthez	2003	40
340	1	Manchester Utd	38	28	5	5	86	43	89	75530	Robin van Persie - 26	David de Gea	2013	43
160	1	Arsenal	38	26	12	0	73	26	90	38079	Thierry Henry - 30	Jens Lehmann	2004	47
320	1	Manchester City	38	28	5	5	93	29	89	47045	Sergio Agüero - 23	Joe Hart	2012	64
180	1	Chelsea	38	29	8	1	72	15	95	41870	Frank Lampard - 13	Petr Čech	2005	57
300	1	Manchester Utd	38	23	11	4	78	37	80	75109	Dimitar Berbatov - 20	Edwin van der Sar	2011	41
280	1	Chelsea	38	27	5	6	103	32	86	41423	Didier Drogba - 29	Petr Čech	2010	71
200	1	Chelsea	38	29	4	5	72	22	91	41902	Frank Lampard - 16	Petr Čech	2006	50
260	1	Manchester Utd	38	28	6	4	68	24	90	75304	Cristiano Ronaldo - 18	Edwin van der Sar	2009	44
220	1	Manchester Utd	38	28	5	5	83	27	89	75826	Cristiano Ronaldo - 17	Edwin van der Sar	2007	56
40	1	Arsenal	38	23	9	6	68	33	78	38053	Dennis Bergkamp - 16	David Seaman	1998	35
460	1	Manchester City	38	32	2	4	95	23	98	54130	Sergio Agüero - 21	Ederson	2019	72
240	1	Manchester Utd	38	27	6	5	80	22	87	75691	Cristiano Ronaldo - 31	Edwin van der Sar	2008	58
480	1	Liverpool	38	32	3	3	85	33	99	41955	Mohamed Salah - 19	Alisson	2020	52
20	1	Manchester Utd	38	21	12	5	76	44	75	55081	Ole Gunnar Solskjær - 18	Peter Schmeichel	1997	32
481	2	Manchester City	38	26	3	9	102	35	81	37097	Raheem Sterling - 20	Ederson	2020	67
161	2	Chelsea	38	24	7	7	67	30	79	41234	Jimmy Floyd Hasselbaink - 13	Carlo Cudicini	2004	37
401	2	Arsenal	38	20	11	7	65	36	71	59944	Olivier Giroud - 16	Petr Čech	2016	29
341	2	Manchester City	38	23	9	6	66	34	78	46974	Edin Džeko - 14	Joe Hart	2013	32
441	2	Manchester Utd	38	25	6	7	68	28	81	74976	Romelu Lukaku - 16	David de Gea	2018	40

In [ ]: