DISTRIBUTIONS IN PYTHON

1) Random variable (RV):-

Maps outcome of a sample space to a real line such that there is a unique real
number corresponding to every outcome of sample space
Eg: Coin toss experiment where outcomes [H,T] are mapped to [0,1].If outcomes are
already real valued then no need to map them on real line (throw of a dice). These
are examples of discrete random variables
Continuous RV maps outcomes of a continuous phenomena to intervals in a real
line (Eg: sensor readings)
2) Probability mass/density function:-

For a discrete RV, the probability mass function assigns a probability measure to
every discrete outcome of sample space
Coin toss sample space [H T] mapped to [0 1]. X is discrete RV whose outcome can
be 0 or 1
P(X = 0) = 0.5, P(X = 1) = 0.5
For a continuous RV, the probability density function assigns a probability measure
to every interval on real line
b
P(a < x < b) = ∫ f(x)dx (Area under the curve of f(x))
a
3) Cumulative distribution/density function:-

This is the probability that the RV 'x' lies in the interval −∞ < x < b
b
F(b) = P(−∞ < x < b) = ∫ f(x)dx
−∞
Every distribution in Python has four functionalities

1. ‘rvs’: generating random numbers from a distribution

2. ‘pdf’:probability density function
3. ‘cdf’:cumulative distribution function
4. ‘ppf’: percentile point function (inverse cumulative distribution function)

Distribution root name

norm- normal distribution
binom- binomial distribution

In this lecture we are going to work with standard normal distributions i.e.
mean=0
standard deviation=1
NORMAL DISTRIBUTION
In [1]: # Importing numpy for numerical operations
import numpy as np

# Importing matplotlib and seaborn for visualization

import matplotlib.pyplot as plt
import seaborn as sns

# Importing scipy.stats for statistical computations

import scipy.stats
 Setting gure size

In [2]: plt.rcParams["figure.figsize"] = (7,7)

 GENERATING RANDOM NUMBERS
In [3]: v1= scipy.stats.norm.rvs(loc=0,scale=1,size=10)
print('Mean',v1.mean())
print('Std Dev',v1.std())

Mean 0.3253569359912848
Std Dev 1.2820175997433112
In [4]: sns.distplot(v1)

Out[4]: <matplotlib.axes._subplots.AxesSubplot at 0x1cfe3a42048>

In [5]: v2= scipy.stats.norm.rvs(loc=0,scale=1,size=100)
print('Mean',v2.mean())
print('Std Dev',v2.std())

Mean 0.020713894103135778
Std Dev 0.9415726759437645

In [6]: sns.distplot(v2)

Out[6]: <matplotlib.axes._subplots.AxesSubplot at 0x1cfe6b01648>

In [7]: v3= scipy.stats.norm.rvs(loc=0,scale=1,size=1000)
print('Mean',v3.mean())
print('Std Dev',v3.std())

Mean 0.02103248207417688
Std Dev 0.9865003013211904

In [8]: sns.distplot(v3)

Out[8]: <matplotlib.axes._subplots.AxesSubplot at 0x1cfe6b87e48>

 PROBABILITY DENSITY FUNCTION
In [9]: scipy.stats.norm.pdf(-1)

Out[9]: 0.24197072451914337
In [10]: scipy.stats.norm.pdf(np.arange(-3,-1,0.01),loc = 0,scale= 1)

Out[10]: array([0.00443185, 0.00456659, 0.00470496, 0.00484703, 0.0049929 ,

0.00514264, 0.00529634, 0.0054541 , 0.00561598, 0.0057821 ,
0.00595253, 0.00612738, 0.00630673, 0.00649068, 0.00667932,
0.00687277, 0.0070711 , 0.00727444, 0.00748287, 0.00769651,
0.00791545, 0.00813981, 0.00836969, 0.0086052 , 0.00884645,
0.00909356, 0.00934664, 0.0096058 , 0.00987115, 0.01014283,
0.01042093, 0.0107056 , 0.01099694, 0.01129507, 0.01160014,
0.01191224, 0.01223153, 0.01255811, 0.01289213, 0.0132337 ,
0.01358297, 0.01394006, 0.01430511, 0.01467825, 0.01505962,
0.01544935, 0.01584758, 0.01625445, 0.0166701 , 0.01709467,
0.0175283 , 0.01797113, 0.01842331, 0.01888498, 0.01935628,
0.01983735, 0.02032836, 0.02082943, 0.02134071, 0.02186237,
0.02239453, 0.02293735, 0.02349099, 0.02405557, 0.02463127,
0.02521822, 0.02581658, 0.02642649, 0.0270481 , 0.02768157,
0.02832704, 0.02898466, 0.02965458, 0.03033696, 0.03103193,
0.03173965, 0.03246027, 0.03319392, 0.03394076, 0.03470094,
0.03547459, 0.03626187, 0.03706291, 0.03787786, 0.03870686,
0.03955004, 0.04040755, 0.04127953, 0.04216611, 0.04306742,
0.0439836 , 0.04491477, 0.04586108, 0.04682264, 0.04779957,
0.04879202, 0.04980009, 0.0508239 , 0.05186358, 0.05291923,
0.05399097, 0.0550789 , 0.05618314, 0.05730379, 0.05844094,
0.05959471, 0.06076517, 0.06195242, 0.06315656, 0.06437766,
0.06561581, 0.06687109, 0.06814357, 0.06943331, 0.07074039,
0.07206487, 0.07340681, 0.07476626, 0.07614327, 0.07753789,
0.07895016, 0.08038011, 0.08182778, 0.08329319, 0.08477636,
0.08627732, 0.08779607, 0.08933262, 0.09088698, 0.09245913,
0.09404908, 0.0956568 , 0.09728227, 0.09892547, 0.10058637,
0.10226492, 0.1039611 , 0.10567483, 0.10740608, 0.10915477,
0.11092083, 0.11270421, 0.1145048 , 0.11632253, 0.1181573 ,
0.120009 , 0.12187754, 0.12376279, 0.12566464, 0.12758295,
0.1295176 , 0.13146843, 0.1334353 , 0.13541806, 0.13741654,
0.13943057, 0.14145997, 0.14350455, 0.14556413, 0.1476385 ,
0.14972747, 0.1518308 , 0.15394829, 0.1560797 , 0.15822479,
0.16038333, 0.16255506, 0.16473972, 0.16693704, 0.16914676,
0.17136859, 0.17360225, 0.17584743, 0.17810384, 0.18037116,
 CUMULATIVE DISTRIBUTION FUNCTION
In [11]: scipy.stats.norm.cdf(x=-1,loc = 0,scale= 1)

Out[11]: 0.15865525393145707

In [12]: 1-scipy.stats.norm.cdf(x=-1,loc = 0,scale= 1)

Out[12]: 0.8413447460685429
VISUALIZING DISTRIBUTIONS
Left Tail Right Tail

Bounded Tails
LEFT TAIL
In [13]: plt.fill_between(x=np.arange(-3,3,0.01),
y1= scipy.stats.norm.pdf(np.arange(-3,3,0.01)),
facecolor='blue',
alpha=0.5)
plt.vlines(x=-1,ymin=0,ymax=0.24,linestyles='dashed')

Out[13]: <matplotlib.collections.LineCollection at 0x1cfe6c49388>

In [14]: scipy.stats.norm.cdf(x=-1,loc = 0,scale= 1)

Out[14]: 0.15865525393145707

cdf to left of -1 i.e. Z<=-1

In [15]: prob_under_neg1 =scipy.stats.norm.cdf(x=-1,loc = 0,scale= 1)
print(prob_under_neg1)

0.15865525393145707
Plotting the cdf value on the normal distribution
In [16]: plt.fill_between(x=np.arange(-3,-1,0.01),
y1= scipy.stats.norm.pdf(np.arange(-3,-1,0.01)) ,
facecolor='blue',edgecolor='black',
alpha=0.5)
plt.fill_between(x=np.arange(-1,3,0.01),
y1= scipy.stats.norm.pdf(np.arange(-1,3,0.01)) ,
facecolor='red',edgecolor='black',
alpha=0.5)
plt.text(x=-2, y=0.03, s= round(prob_under_neg1,3))
plt.vlines(x=-1,ymin=0,ymax=0.24,linestyles='dashed')

Out[16]: <matplotlib.collections.LineCollection at 0x1cfe6d08e88>

RIGHT TAIL
In [17]: plt.fill_between(x=np.arange(-3,3,0.01),
y1= scipy.stats.norm.pdf(np.arange(-3,3,0.01)) ,
facecolor='blue',
alpha=0.5)
plt.vlines(x=-1,ymin=0,ymax=0.24,linestyles='dashed')

Out[17]: <matplotlib.collections.LineCollection at 0x1cfe6cce408>

 cdf to right of -1 i.e. Z>=-1
In [18]: prob_over_neg1 =1-prob_under_neg1
print(prob_over_neg1)

0.8413447460685429
In [19]: plt.fill_between(x=np.arange(-3,-1,0.01),
y1= scipy.stats.norm.pdf(np.arange(-3,-1,0.01)) ,
facecolor='red',edgecolor='black',
alpha=0.5)
plt.fill_between(x=np.arange(-1,3,0.01),
y1= scipy.stats.norm.pdf(np.arange(-1,3,0.01)) ,
facecolor='blue',edgecolor='black',
alpha=0.5)
plt.text(x=0.5, y=0.03, s= round(prob_over_neg1,3))
plt.vlines(x=-1,ymin=0,ymax=0.24,linestyles='dashed')

Out[19]: <matplotlib.collections.LineCollection at 0x1cfe6dffdc8>

BOUNDED
In [20]: plt.rcParams["figure.figsize"] = (5,5)
plt.fill_between(x=np.arange(-3,-1,0.01),
y1= scipy.stats.norm.pdf(np.arange(-3,-1,0.01)) ,
facecolor='red',edgecolor='black',
alpha=0.5)
plt.fill_between(x=np.arange(-1,1,0.01),
y1= scipy.stats.norm.pdf(np.arange(-1,1,0.01)) ,
facecolor='blue',edgecolor='black',
alpha=0.5)
plt.fill_between(x=np.arange(1,3,0.01),
y1= scipy.stats.norm.pdf(np.arange(1,3,0.01)) ,
facecolor='red',edgecolor='black',
alpha=0.5)
plt.vlines(x=-1,ymin=0,ymax=0.24,linestyles='dashed')
plt.vlines(x=1,ymin=0,ymax=0.24,linestyles='dashed')

Out[20]: <matplotlib.collections.LineCollection at 0x1cfe6e836c8>

In [21]: prob_over_pos1 =1-scipy.stats.norm.cdf(x=1,loc = 0,scale= 1)
print(prob_over_pos1)
print(prob_under_neg1)
between_prob=1-(prob_under_neg1+prob_over_pos1)
print(between_prob)

0.15865525393145707
0.15865525393145707
0.6826894921370859
 TAILS
In [22]: tails_prob=prob_under_neg1+prob_over_pos1
print(tails_prob)

0.31731050786291415
INVERSE CUMULATIVE DISTRIBUTION FUNCTION
In [23]: plt.fill_between(x=np.arange(-3,3,0.01),
y1= scipy.stats.norm.pdf(np.arange(-3,3,0.01)) ,
facecolor='blue',
alpha=0.5)

plt.vlines(x=-1,ymin=0,ymax=0.24,linestyles='dashed')

Out[23]: <matplotlib.collections.LineCollection at 0x1cfe6e3fc88>

In [24]: q_val=scipy.stats.norm.cdf(x=-1,loc = 0,scale= 1)
print(q_val)

0.15865525393145707

In [25]: scipy.stats.norm.ppf(q=q_val,loc = 0,scale= 1)

Out[25]: -1.0
 BINOMIAL DISTRIBUTION
No. of successes in 10 tosses of a fair coin
n- no.of tosses
p- probability of success

In [26]: scipy.stats.binom.rvs(n=10,p=0.5)

Out[26]: 5
 A coin is tossed 10 times and the experiment is repeated for 5 times.In each experiment
the number of successes are recorded

In [27]: scipy.stats.binom.rvs(size=5,n=10,p=0.5,random_state=0)

Out[27]: array([5, 6, 5, 5, 5])

 Now let us repeat the experiment for 15 times

In [28]: scipy.stats.binom.rvs(size=15,n=10,p=0.5,random_state=0)

Out[28]: array([5, 6, 5, 5, 5, 6, 5, 7, 8, 5, 6, 5, 5, 7, 3])

 Let us visualize this distribution
In [29]: ax=sns.distplot(scipy.stats.binom.rvs(size=15,n=10,p=0.5,random_state=0),kde=False,hist_
kws={"color": 'b','alpha':0.8})
# default color is blue

# Labels for x and y axis

ax.set(xlabel='No. of successes',ylabel='Frequency of success in a given trial')

Out[29]: [Text(0, 0.5, 'Frequency of success in a given trial'),

Text(0.5, 0, 'No. of successes')]
In [30]: ax=sns.distplot(scipy.stats.binom.rvs(size=15,n=10,p=0.5,random_state=0),kde=True,hist_k
ws={"color": 'b','alpha':0.8})
# default color is blue

# Labels for x and y axis

ax.set(xlabel='No. of successes',ylabel='Frequency of success in a given trial')

Out[30]: [Text(0, 0.5, 'Frequency of success in a given trial'),

Text(0.5, 0, 'No. of successes')]
 Now increase the no. of experiments to 100 and then to 500
In [31]: ax=sns.distplot(scipy.stats.binom.rvs(size=100,n=10,p=0.5,random_state=0),kde=False,hist
_kws={"color": 'b','alpha':0.8})
# default color is blue

# Labels for x and y axis

ax.set(xlabel='No. of successes',ylabel='Frequency of success in a given trial')

Out[31]: [Text(0, 0.5, 'Frequency of success in a given trial'),

Text(0.5, 0, 'No. of successes')]
In [32]: ax=sns.distplot(scipy.stats.binom.rvs(size=100,n=10,p=0.5,random_state=0),kde=True,hist_
kws={"color": 'b','alpha':0.8})
# default color is blue

# Labels for x and y axis

ax.set(xlabel='No. of successes',ylabel='Frequency of success in a given trial')

Out[32]: [Text(0, 0.5, 'Frequency of success in a given trial'),

Text(0.5, 0, 'No. of successes')]
In [33]: ax=sns.distplot(scipy.stats.binom.rvs(size=500,n=10,p=0.5,random_state=0),kde=True,hist_
kws={"color": 'b','alpha':0.8})
# default color is blue

# Labels for x and y axis

ax.set(xlabel='No. of successes',ylabel='Frequency of success in a given trial')

Out[33]: [Text(0, 0.5, 'Frequency of success in a given trial'),

Text(0.5, 0, 'No. of successes')]
 PROBABILITY MASS FUNCTION
Probability of seeing a given number of success

Probability of seeing exactly nine heads i.e.X=9

In [34]: scipy.stats.binom.pmf(n=20,p=0.5,k=9)

Out[34]: 0.16017913818359344
 Probability of seeing nine or lesser heads i.e. X<=9

In [35]: scipy.stats.binom.pmf(n=20,p=0.5,k=0)+\
scipy.stats.binom.pmf(n=20,p=0.5,k=1)+\
scipy.stats.binom.pmf(n=20,p=0.5,k=2)+\
scipy.stats.binom.pmf(n=20,p=0.5,k=3)+\
scipy.stats.binom.pmf(n=20,p=0.5,k=4)+\
scipy.stats.binom.pmf(n=20,p=0.5,k=5)+\
scipy.stats.binom.pmf(n=20,p=0.5,k=6)+\
scipy.stats.binom.pmf(n=20,p=0.5,k=7)+\
scipy.stats.binom.pmf(n=20,p=0.5,k=8)+\
scipy.stats.binom.pmf(n=20,p=0.5,k=9)

Out[35]: 0.4119014739990231

In [36]: k_range=np.arange(0,10)

scipy.stats.binom.pmf(n=20,p=0.5,k=k_range)

sum(scipy.stats.binom.pmf(n=20,p=0.5,k=k_range))

Out[36]: 0.4119014739990231
 This is the cumulative distribution function
In [37]: scipy.stats.binom.cdf(n=20,p=0.5,k=9)

Out[37]: 0.41190147399902316