Consulting - Quality
Difference makes the DIFFERENCE
NUMPY EXERCISE
from PadhAI homework
In [3]:
import numpy as np
In [4]:
scores = np.loadtxt("/content/cric_data.tsv", skiprows = 1)
split dataset based on player¶
In [133]:
sac = np.loadtxt("/content/cric_data.tsv", skiprows = 1, usecols = 1)
In [6]:
rah = np.loadtxt("/content/cric_data.tsv", skiprows = 1, usecols = 2)
In [7]:
ind = np.loadtxt("/content/cric_data.tsv", skiprows = 1, usecols = 3)
In [ ]:
scores
In [ ]:
np.min(scores)
Out[ ]:
In [ ]:
np.mean(scores)
Out[ ]:
In [ ]:
np.median(scores)
Out[ ]:
Inter quartile range for datasets¶
In [ ]:
iqr = np.percentile(scores, 75) - np.percentile(scores, 25)
In [ ]:
print(iqr)
In [ ]:
iqr = np.percentile(scores, [75, 25])
In [ ]:
print(iqr)
In [ ]:
print(iqr[0] - iqr[1])
In [ ]:
np.mean(sac)
Out[ ]:
In [ ]:
np.mean(rah)
Out[ ]:
In [ ]:
np.mean(ind)
Out[ ]:
In [ ]:
sac_iqr = np.percentile(sac, [75, 25])
In [ ]:
sach_iqr = sac_iqr[0] - sac_iqr[1]
In [ ]:
print(sach_iqr)
In [ ]:
np.median(sac)
Out[ ]:
In [ ]:
print(np.percentile(sac, 50))
In [ ]:
print(np.mean(rah))
In [ ]:
print(np.median(rah))
In [ ]:
rah_iqrs = np.percentile(rah, [75, 25])
In [ ]:
print(rah_iqrs[0] - rah_iqrs[1])
Q2. histogram for India dataset¶
In [ ]:
np.histogram(sac, 10)
Out[ ]:
In [ ]:
np.histogram(rah, 4)
Out[ ]:
In [ ]:
np.histogram(ind, 20)
Out[ ]:
In [ ]:
np.split?
Q3. split array into groups (in this case - 25) on sacs data¶
In [ ]:
np.split(sac, 5)
Out[ ]:
In [ ]:
sac25grp = np.split(sac, 25)
In [ ]:
sac25grp
Out[ ]:
In [ ]:
np.median(sac25grp[0])
Out[ ]:
In [ ]:
np.median(sac25grp[1])
Out[ ]:
In [ ]:
print(np.mean(sac25grp))
In [ ]:
print(np.mean(sac25grp[0:24]))
to get mean for each element in the array that is split into 25 subgroups - without traditional for loop¶
In [ ]:
sac25grp_mean = [np.mean(i) for i in sac25grp]
In [ ]:
sac25grp_mean
Out[ ]:
In [ ]:
print([np.mean(i) for i in sac25grp])
In [ ]:
print([np.median(i) for i in sac25grp])
In [ ]:
rah25grp = np.split(rah, 25)
In [ ]:
rah25grp
Out[ ]:
In [ ]:
rah25grp_median = [np.median(i) for i in rah25grp]
In [ ]:
rah25grp_median
Out[ ]:
In [ ]:
sac
Out[ ]:
In [ ]:
sacgr100 = [sac[i] for i in sac if sac>100]
In [ ]:
sacgr100
In [ ]:
sacgr100 = sac > 100
In [ ]:
sacgr100
In [ ]:
from PADHAI solutions video¶
Dataset for the exercise questions¶
Q1. Find the mean, median, IQR for Sachin, Rahul and India¶
In [8]:
import numpy as np
In [9]:
cric_data = np.loadtxt("/content/cric_data.tsv", skiprows = 1)
In [10]:
cric_data.shape
Out[10]:
In [11]:
cric_data = cric_data[:, [1,2,3]]
In [13]:
cric_data
Out[13]:
In [14]:
sac = cric_data[:,0]
In [135]:
sac
Out[135]:
In [16]:
rah = cric_data[:,1]
In [17]:
rah
Out[17]:
In [18]:
india = cric_data[:, 2]
In [19]:
india
Out[19]:
In [20]:
def stats(col):
print("Mean is : ", np.mean(col))
print("Median is :", np.median(col))
print("Inter quartile range :", np.percentile(col, 75)
- np.percentile(col, 25))
In [21]:
stats(sac)
In [22]:
stats(rah)
In [23]:
stats(india)
In [24]:
np.mean(cric_data, axis = 0)
Out[24]:
In [25]:
np.median(cric_data, axis=0)
Out[25]:
In [26]:
np.percentile(cric_data, 75, axis=0) - np.percentile(cric_data, 25, axis = 0)
Out[26]:
Q2. Find the histogram of Sachin's scores with 10 bins¶
In [27]:
np.histogram(sac)
Out[27]:
Q3. Find the mean of Sachin's scores grouped by 25 matches¶
In [28]:
sac.shape
Out[28]:
Considering sachin's dataset, when reshape is applied, the first 25 datapoints form the first row, the second 25 data poits forms the second row and so on...¶
In [29]:
sac.reshape(9, 25)
Out[29]:
In [30]:
sac_by_25 = sac.reshape(9, 25)
In [31]:
np.mean(sac_by_25, axis = 1)
Out[31]:
In [32]:
sac > 100
Out[32]:
In [33]:
sac[sac>100]
Out[33]:
In [34]:
np.mean(sac[sac > 100])
Out[34]:
In [35]:
## Question 5
np.mean(sac[rah <= 10])
Out[35]:
Q6. Find mean for Sachin's scores based on which quartile India's score falls in¶
In [36]:
np.percentile(india, [25, 50, 75, 100])
Out[36]:
In [37]:
qrs = np.percentile(india, [25, 50, 75, 100])
In [38]:
india
Out[38]:
In [39]:
india < 175
Out[39]:
In [40]:
india[(india < 175)]
Out[40]:
In [41]:
india[india < [175, 215]]
In [43]:
qrs
Out[43]:
In [44]:
india < qrs
In [82]:
qrs.shape
Out[82]:
Useful link to BROADCASTING: https://numpy.org/doc/stable/user/basics.broadcasting.html¶
General Rule(s):
- are they equal, or
- one of them is 1
In [83]:
india.shape
Out[83]:
In [85]:
qrs = qrs.reshape(4, 1)
In [86]:
qrs.shape
Out[86]:
In [87]:
india < qrs
Out[87]:
In [88]:
indicies = india < qrs
In [89]:
indicies.shape
Out[89]:
In [90]:
qrs
Out[90]:
In [91]:
sac
Out[91]:
dataset for scores less than 216 - which is the second quartile¶
In [92]:
sac[indicies[1, :]]
Out[92]:
dataset for scores less than 175 - which is the first quartile¶
In [93]:
sac[indicies[0,:]]
Out[93]:
one liner for FOR Loop - to get mean scores for all the four quartiles¶
In [94]:
for i in range(4): print(i, np.mean(sac[indicies[i]]))
Q7. For every match find out who has scored more - Sachin or Dravid¶
In [95]:
snr = cric_data[0 : 2]
In [96]:
snr
Out[96]:
In [97]:
snr = cric_data[:, 0:2]
In [98]:
snr
Out[98]:
using argmax f() - returns the position of the max value in the given array¶
In [99]:
np.max([1,2,3,4,3,2])
Out[99]:
In [100]:
np.argmax([1,2,3,4,3,2])
Out[100]:
In [101]:
np.argmax(snr, axis = 1)
Out[101]:
In [102]:
is_rahul = np.argmax(snr, axis=1)
In [103]:
is_rahul
Out[103]:
In [104]:
np.sum(is_rahul) / 225
Out[104]:
where f() - takes three arguments - condition, x, y
- if true - returns x else
- returns y
In [105]:
np.where(is_rahul, "rahul", "sac")
Out[105]:
Q8. How many more runs do sachin score on average after having scored x runs¶
In [106]:
x_arr = np.arange(0, 101, 5)
In [107]:
x_arr.shape
Out[107]:
In [108]:
x_arr
Out[108]:
In [121]:
x_arr = x_arr.reshape(x_arr.shape[0], 1)
In [110]:
x_arr
Out[110]:
In [111]:
x_arr.shape
Out[111]:
In [112]:
sac.shape
Out[112]:
In [113]:
sac > x_arr
Out[113]:
In [114]:
indicies = sac>x_arr
In [115]:
indicies
Out[115]:
In [116]:
indicies.shape
Out[116]:
In [123]:
# to get all the scores that are more than 5
sac[indicies[1,:]]
Out[123]:
In [124]:
# to get all the scores more than 10
sac[indicies[2, :]]
Out[124]:
In [127]:
# to get scores that are more than 15 and likewise...
sac[indicies[3,:]]
Out[127]:
In [128]:
np.mean(sac[indicies[1,:]])
Out[128]:
In [129]:
np.mean(sac[indicies[2, :]])
Out[129]:
In [130]:
np.mean(sac[indicies[3, :]])
Out[130]:
In [118]:
# the following code gives us the mean of the scores
for i in range(x_arr.shape[0]):
print(np.mean(sac[indicies[i, :]]))
In [136]:
# we are interested in the incremental scores rather than mean so,
# subtract the ith element (what he has already scored) from the mean
for i in range(x_arr.shape[0]):
print(x_arr[i, 0],
np.mean(sac[indicies[i, :]]) - x_arr[i, 0])
Q9. How many matches did Sac take to score first 1000 runs, and next 1000 runs and so on...¶
In [137]:
sac_cumsum = np.cumsum(sac)
In [138]:
sac_cumsum
Out[138]:
In [142]:
np.histogram(sac_cumsum, bins = np.arange(0, 10000, 1000))
Out[142]:
Test¶
In [143]:
a = np.array([1,2,3,4])
print(a[::-1])
In [146]:
a = np.array([1,2,3,4])
a[0] = 0.9
print(a[0], a.dtype)
In [145]:
a
Out[145]:
In [147]:
a = np.arange(3)
b = a.reshape(2,2)
print(b)
In [149]:
a = np.array([1,2,3,4, 5])
print(a[2:None])
In [150]:
a = np.arange(10)
b = np.where(a % 2 == 1, 1, a)
print(b)
In [151]:
a = np.array([[1,2,3], [4,5,6]])
print(a[1][2])
In [153]:
a = np.array([3,6,9,12,15])
b = np.array([2,3,4,5,12])
c = np.where(a == b)
print(c[0])
In [ ]: