NumPy

• stands for Numerical Python
• Designed for scientific computing and analysis
• Deals with multidimensional arrays and matrices.
• Includes a large collection of high level mathematical functions
• Faster than standard python packages

Applications

• Images
• Audio
• Measurements
• Numbers
• Libraries
  • pandas
  • SciPy
  • OpenCV

pip install numpy
pip3 install numpy

NumPy Arrays

• Main data structure in Numpy
• can be single of mutiudimensional
• Contains elements of a single type

Numpy Arrays versus Python List

• No loop structure required for iteration
• Allows only one type per array
  • less flexible
  • makes type checking unnecessary
  • faster performance

NumPy Array Efficiency

• Memory usage is much smaller with arrays
• In lists, each item is pointer (occupying between 4 to 16 bytes)
• NumPy arryas are contiguous blocks of memory, so no need for pointers
  • all items are fixed size and of one type

np . array Constructor

np.array(object, dtype)

# return a new NumPy array from a given object
# object can be any array like object
# dtype is the type of the array
# if not specified, it will be set automatically based on contents of the object

Integer Types

np.bool	boolean
np.int8	byte (-128-127)
np.int16	16bit-signed-integere (-2hoch15 to 2 hoch 15 -1)
np.int32	32bit-signed-integere (-2hoch31 to 2 hoch 31 -1)
np.int64	64bit-signed-integere (-2hoch63 to 2 hoch 63 -1)

Unsigned Types

np.uint8	byte (0 - 256)
np.uint16	16bit-signed-integere (0 to 2 hoch 16)
np.uint32	32bit-signed-integere (0 to 2 hoch 32)
np.uint64	64bit-signed-integere (0 to 2 hoch 64)

Float/Complex Types

np.float32	float
np.float64	double
np.complex64	float complex
np.complex128	double complex

One Dimensional Array Creation

a1 = np.array([1.2,32.4,7.5,3.4])
# one dimensional array from a list of floats
# np.array selcts a type based on the input we have given it

Two Dimensional Array Creation

a2 = np.array([2,4,10,15,13], [5,3,2,13,7],dtype=np.int16 )
# the argument is a list of lists, and the list of lists has two lists, so it makes a 
# two dimensional array 

Array Population Methods

• Fill an array with:
  • zeros, ones or any specific number
  • a range of numbers
  • a range of random numbers

Operators

• Operations can be performed on scalars or arrays with the same shape
• Arithmetic operators (+- * ** / // %)
• Comparison Operators (< <= == != => >)

Arithmetic Operators Example

a = np.array([2,10,-4, 5, 3.7, 80])
print(a + 2)
print(a * 10-5)
# Output [ 4. 12. -2. 7. 5.7 82. ]
			 [ 15. 95. -45. 45. 32. 795. ]
# we are not using a for loop

Comparison Operators Example

• comparing an array with a scalar

a = np.array([2, 10, -4, 5, 3.7, 80])
print(a == 10)
print(a >= 5)
# Output [ False True False False False False ]
			 [ False True False True False True ]
# we are not using a for loop

Other Methods

• sum, product
• minimum, maximum
• Median, average, mean and standard deviaation
• Histogram

---

Recommendations

• numpy.org
  • interactive shell to try NumPy in the browser
• matplotlib.org
  • creating static, animated, and interactive visualizations
  • has tons of Plot types (in py or Jupyter)
• pandas.pydata.org
  • pip install pandas
  • conda install -c conda-forge pandas
  • user guide
• SQL tutorial