1. Data Analysis - NumPy Operations

 Numpy Operations  

Numpy Array Creation - 

• X = np.array[[10,20,30,40]]
• X = np.zeros(10)
• X= np.ones(10)
• X = np.empty (10)
• X=np.arange([1,11])  ---- will create 10 element array starting from element=1 to 10

Array Creation using data type

• X = np.array([10,20,30,40], dtype = np.floar64)

Changing data type

• X = X.astype(np.int32)

Arithmetic operation

• X = np.array([[10,20,30,40],[50,60.70,80])

Indexing and slicing 

• One dimension 
• X[1] ----- Will return X[1]
• X[2:4] ------ will return X[2[, X[3]
• 
  
• Multiple dimension
• X[0] ------- Will return 0th row
• X[0][1] -- Will return 0th row and 1st column.

Propagation and Broadcast

• X[2:4] = 100   ---------- X[2], X[3] will be assigned 100.

Row slicing & Column slicing using booleans

• Y = np.array(["True", "True", "False", "False", "True"]),     
• X[Y=="True"]      -- row slicing
• X[:, Y="True"]     -- column slicing

Manipulating all elements

• X = X[X<0] = 0  --- Any element less than 0 will be assigned a value = 0

Fancy Indexing

• Get rows
• X[1]
• X[[1,3,5]]
• 
  
• Get Columns
• X[:, 1:4]
• X[:, [1,4,5]] 
• 
  
• Get Multiple rows and columns
• X[[1,3,4], [2,4,5]]

Fancy Indexing using pipe

• X[[1,4,5]][:, [0,1]]

Reshape

• X.reshape(8,4)
• X.transpose()

Universal Function

• One input
• np.sqrt(X)
• np.exp(X)
• Multiple Input
• np.maximum(X,Y)
• np.add(X,Y)
• Finding NAN value 
• np.nan(X)  -- Ouput will be [False, False,..., True] -- in this form. -- TRUE will indicate corresponding value in NAN.
• Any element has NAN value
• np.nan(X).any()

Vectorization

• Mesh Grid
• Xaxis, Yaxis = np.meshgrid(X,Y)
• np.where
• K = np.where(Z, X, Y)
• K = np.where(X>0, Y ,Z )
• np.where(X>99, X, 100) -> combination of scaler and numpy array

Statistical Functions

• np.mean(X) --- Mean of all element of the array.
• np.std(X)
• np.sum()
• X.mean(axis=0) -- Mean of each column
• X.mean(axis=1) -- Mean of each row.

Boolean arrays

• X.any() - IF any value is TRUE
• X.sum() - Count of all TRUE value
• X.all() - If all value is TRUE
• X>0 -> returns a boolean array.

Sorting

• np.sort(X, axis = None) -- sort all element of array and return one dimensional array
• np.sort(X, axis = 0) - Column wise sorting
• np.sort(X, axis = 1 ) - Row wise sorting

Remove duplicates

• np.unique(X) --  removes the duplicates

Check common values between two array

• np.in1d(X,Y) -.. returns bollean array with TRUE indicating common value
• np.union1d(X,Y) - returns union of X and Y

Saving to a file

Saving one array in one file

• np.save("OneDimension", X)
• np.load("OneDimension.npy")

Save multiple array using key/value

• np.savez("Onedimension", key1=X, key2=Y)
• k = np.load("Onedimension")
• k[key1], k[key2]

Other functionality

• np.dot(X,Y)
• k= inv(y) ---- from numpy.linalg import inv
•