#!/usr/bin/env python
# coding: utf-8

# In[1]:


import numpy as np


# ### Arrays

# In[2]:


# numpy array
arr = np.array([0,1,2,3])
arr


# numpy calc is faster than native python 

# In[3]:


# testing the same

L = range(100000)
get_ipython().run_line_magic('timeit', '[i**2 for i in L]')


# In[4]:


L = np.arange(100000)
get_ipython().run_line_magic('timeit', 'L**2')


# there are so many ways to create array in numpy, we will see them one by one
# #### starting with 1-D array

# In[5]:


a = np.array([1,2,5,3,5,9])
a


# In[6]:


a = np.arange(1, 20)    # start, end
a


# In[7]:


a = np.arange(1, 20, 3)   # start, end, step
a


# In[8]:


# check dimension and size
print("dimension - ", a.ndim)
print("size - ", a.shape)
print("length - ", len(a))


# #### 2-D array

# In[9]:


b = np.array([[1,2,3], [2,3,4], [3,4,5]])
b


# In[10]:


# check dimension and size
print("dimension - ", b.ndim)
print("size - ", b.shape)
print("length - ", len(b))


# 3-D array

# In[11]:


c = np.array([[[5,7], [8,9], [0,7]], [[1,2], [2,3], [6,7]], [[5,7], [8,9], [0,7]]])
c


# In[12]:


# check dimension and size
print("dimension - ", c.ndim)
print("size - ", c.shape)
print("length - ", len(c))


# Simplest way to create test array

# In[13]:


# by no
a = np.arange(1, 10, 2)
print(a)
print(a.shape)


# In[14]:


# by number of points
a = np.linspace(1, 10, 4)  # pick any 4 no from 1 to 10
print(a)
print(a.shape)


# In[15]:


# by number of points
a = np.linspace(1, 10, 6, endpoint=False)  # pick any 4 no from 1 to 10
print(a)
print(a.shape)


# Common arrays to use :

# In[16]:


np.ones(4)


# In[17]:


np.ones((2,4))


# In[18]:


np.zeros((2,3))


# In[19]:


np.eye((3))


# In[20]:


np.diag([1,2,3,4])


# #### Random no generation 
# with help of numpy random library

# In[21]:


np.random.rand(5)   # pick any 5 value between 0 and 1


# In[22]:


np.random.randn(5)  # includes negative no as well (Gaussian)


# In[23]:


np.random.randint(9)  # this will pick any random no from 0 to 8 (integer)


# In[24]:


np.random.randint(10, 13)  # start, end    end exclusive


# In[25]:


np.random.randint(10, 13, 2)  # start, end, no of element    end exclusive


# In[26]:


np.random.randint(10, 13, 4)  # start, end, no of element    end exclusive


# In[27]:


np.random.randint(1, 20, (4, 3))


# If you set the np.random.seed(a_fixed_number) every time you call the numpy's other random function, the result will be the same:

# In[28]:


np.random.seed(123)
np.random.randint(1, 10, 2)


# In[29]:


np.random.seed(123)
np.random.randint(1, 10, 2)


# #### Data Type

# In[30]:


a = np.random.randint(3,9, 3)
print(a)
print(a.dtype)


# In[31]:


a = np.random.rand(3)
print(a)
print(a.dtype)


# You can explicitly specify which data-type you want:

# In[32]:


c = np.array([1, 2, 3], dtype=float)
c.dtype


# In[33]:


c = np.array([1, 2, 3])
c.dtype


# In[34]:


a = np.ones((3, 3))
a.dtype


# In[35]:


# complex data type
d = np.array([1+2j, 3+4j, 5+6*1j])
d.dtype


# In[36]:


# Boolean
e = np.array([True, False, False, True])
e.dtype


# In[37]:


# String
f = np.array(['Bonjour', 'Hello', 'Hallo'])
f.dtype  # unicode string with max 7 letters


# ### Indexing and Slicing

# In[38]:


a = np.arange(10)
print(a)

a[2], a[0], a[8], a[-1]


# In[39]:


a[2:7]


# In[40]:


# reversing the array
a[::-1]


# In[41]:


a[1:8:3]  #start, end, step


# In[42]:


a[:4]  #start, 4


# In[43]:


a[6:]  #6, end


# In[44]:


a[:]  #start, end


# numpy array supports all python Indexing and Slicing operation
# #### Numpy indexing and Slicing

# In[45]:


a = np.arange(10)
a


# In[46]:


a[5:] = 11
a


# In[47]:


a[6:] = a[::-1][6:]
a


# ### Array Copies and Views

# In[48]:


a = np.arange(10)
a


# In[49]:


b = a[6:]
b


# In[50]:


a[7] =  13
print(a)
print(b)


# In[51]:


print(np.may_share_memory(a,b))


# #### Copy

# In[52]:


b = a[6:].copy()
b


# In[53]:


a[8] =  16
print(a)
print(b)


# In[54]:


print(np.may_share_memory(a,b))


# #### Boolean indexing or Fancy indexing

# In[55]:


a = np.arange(10)
a


# In[56]:


b = a%3==0
b   # boolean array


# In[57]:


# picking only those for which a%3==0 is TRUE
a[b]   # boolean indexing


# In[58]:


a[b] = a[b]+200
a


# In[59]:


b = np.random.randint(0, 6, 3)
b


# In[60]:


a[b]


# In[ ]: