Multi-dimensional array objects ndarray and numpyndarray

Multi-dimensional array objects ndarray and numpyndarray

PS: content comes from "Data Analysis Using Python"

 

1. Create an ndarray

1. array: converts a sequence (nested sequence) into an array (multi-dimensional array)

In[2]: import numpy as npIn[3]: arr = np.array([1, 2, 3, 4, 5, 6, 7, 8, 9])In[4]: arrOut[4]: array([1, 2, 3, 4, 5, 6, 7, 8, 9])In[5]: arr = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])In[6]: arrOut[6]: array([[1, 2, 3],       [4, 5, 6],       [7, 8, 9]])

View Code

2. arange: similar to the built-in range, an ndarray object is returned. Specify array dimensions through reshape

In[8]: arr = np.arange(1, 10).reshape(3, 3)In[9]: arrOut[9]: array([[1, 2, 3],       [4, 5, 6],       [7, 8, 9]])

View Code

3. ones, ones_like, zeros, and zeros_like: Create an array of all 1 or all 0 based on the specified shape and dtype. The ones_like or zeros_like array is a parameter, create an array of all 1 or all 0 based on the shape and dtype of the parameter Array

arr = np.ones((2, 4), dtype = np.int32)arrOut[11]: array([[1, 1, 1, 1],       [1, 1, 1, 1]])arr2 = np.zeros_like(arr)arr2Out[13]: array([[0, 0, 0, 0],       [0, 0, 0, 0]])

View Code

4. empty and empy_like: similar to ones and ones_like, but the allocated memory space is not filled with any value, that is, the elements of the array are not initialized.

arr = np.empty((3, 3), dtype = np.float64)arrOut[15]: array([[ 0.,  0.,  0.],       [ 0.,  0.,  0.],       [ 0.,  0.,  0.]])arr1 = np.empty_like(arr)arr1Out[17]: array([[ 0.,  0.,  0.],       [ 0.,  0.,  0.],       [ 0.,  0.,  0.]])

View Code

5. eye and identity: Create an N x N matrix (the diagonal line is 1, and the rest is 0)

In[18]: arr = np.eye(4, dtype = np.float32)In[19]: arrOut[19]: array([[ 1., 0., 0., 0.],[ 0., 1., 0., 0.],[ 0., 0., 1., 0.],[ 0., 0., 0., 1.]], dtype=float32)In[20]: arr1 = np.identity(4, dtype = np.int32)In[21]: arr1Out[21]: array([[1, 0, 0, 0],[0, 1, 0, 0],[0, 0, 1, 0],[0, 0, 0, 1]])

View Code

 

2. shape and dtype of ndarray

Ndarray is a universal multi-dimensional container of homogeneous data. All elements must be of the same data type.

Ndarray. shape gets a tuple that represents the size of each dimension of the array. Ndarray. dtype: obtain the data type of the array element.

Numpy data types include: int8 int16 int32 int64 uint8 uint16 uint16 uint64 float16 float32 float64 float128 complex64 complex128 complex256 bool object (O) string _ (S) unicode _ (U)

In[2]: import numpy as npIn[3]: arr = np.arange(16).reshape((2, 2, 4))In[4]: arr.shapeOut[4]: (2, 2, 4)In[5]: arr.dtypeOut[5]: dtype('int32')

View Code

You can use the astype method of ndarray to display and convert dtype. If the conversion process fails (the type conversion cannot be implemented), a TypeError is thrown. Calling astype creates a new array, even if the new and old dtpye are the same.

In[16]: arr = np.array(10)In[17]: arr.dtypeOut[17]: dtype('int32')In[18]: float_arr = arr.astype(np.float32)In[19]: float_arr.dtypeOut[19]: dtype('float32')In[20]: str = np.array(['1.3', '-4.5', '33'], dtype = np.string_)In[21]: str.dtypeOut[21]: dtype('S4')In[22]: num = str.astype(np.float32)In[23]: num.dtypeOut[23]: dtype('float32')

View Code

 

Iii. vector operation between arrays and scalar

Any arithmetic operation between arrays of the same size applies the operation to the element level.

In[26]: arr = np.arange(1, 10).reshape((3, 3))In[27]: arrOut[27]: array([[1, 2, 3],       [4, 5, 6],       [7, 8, 9]])In[28]: arr ** 2Out[28]: array([[ 1,  4,  9],       [16, 25, 36],       [49, 64, 81]])In[29]: arr - arrOut[29]: array([[0, 0, 0],       [0, 0, 0],       [0, 0, 0]])In[30]: arr * arr.TOut[30]: array([[ 1,  8, 21],       [ 8, 25, 48],       [21, 48, 81]])

View Code