Numpy Cheat Sheet

ARRAYS

An ordered collection of elements that can be modified (mutable) and retrieved (indexed). This collection may be multi-dimensional.

Array creation

myArray = numpy.array(['abc', 'bca', 'cab'])	Create a new one-dimensional array called ‘myArray’ with string elements.
my2DArray = numpy.array([[0, 15, 4],[8, 12, 0]])	Create a new two-dimensional array called my2DArray with integer elements
newArr = numpy.array(variable)	Convert variable into a numpy array
newEmptyArr = numpy.empty((shape1, shape2), dtype=np.float32)	Create a new two-dimensional array that is empty but will store floats shape1*shape2 floats where shape1 is number of rows
newZerosArr = numpy.zeros((shape1, shape2))	Create a new two-dimensional array that contnains only zeros, where shape1 is the number of rows
newFullArr = numpy.full((shape1, shape2), value)	Create a new two-dimensional array that contains some value value, like numpy.nan or 42; shape1 is the number of rows

Subsetting and data-selection

	retrieved = myArray[index]	Retrieve an item from the one-dimensional myArray at the given index. You can change the item at the given index by setting it equal to a new value. An index of -1 will return the last item in the list. Indexing begins at 0
	retrieved = my2DArray[index]	Retrieve a row of elements at the given index (dtype is another numpy array) from the two-dimensional my2DArray. Indexing begins at 0
	retreived = my2DArray[:, index]	Retrieve a column of elements at the given index (dtype is another numpy array) from the two-dimensionnal my2DArray. Indexing begins at 0
	retrieved = my2DArray[index1, index2]	Retrieve a single element at the given indices from the two-dimensional my2DArray. Indexing begins at 0
	slicedArr = myArray[start:stop]	Create a new array from a slice of myArray. Either start or stop can be ommitted (e.g. myArray[:3] will give you a slice of myArray up to index 3). Indexing again begins at 0

Array attributes

myArray.ndim	Return the number of dimensions of myArray
myArray.shape	Return the shape (or number of elements) of each dimension of myArray as a tuple. (e.g. myArray.shape[0] will give you the shape of the first dimension)

Operations and inspecting data

my2DArray += value	Add an int or float value to every element in the array my2DArray
min_value = numpy.amin(my2DArray, axis=0)	Return the minimum value for each column in the 2D array
min_value2 = numpy.amin(my2DArray, axis=1)	Return the minimum value for each row in the 2D array
min_value3 = numpy.amin(my2DArray)	Return the minimum value in the whole 2D array
max_value = numpy.amax(my2DArray, axis=0)	Return the maximum value for each column in the 2D array
max_value2 = numpy.amax(my2DArray, axis=1)	Return the maximum value for each row in the 2D array
max_value3 = numpy.amax(my2DArray)	Return the maximum value in the whole 2D array
min_val = numpy.minimum(my2DArray, value)	Returns the minimum value element-wise, comparing my2DArray and the given value, or another array of the same shape. If any NaNs are present, that is the default minimum
max_val = numpy.maximum(my2DArray, value)	Returns the maximum value element-wise, comparing my2DArray and the given value, or another array of the same shape. If any NaNs are present, that is the default maximum
indices_true = numpy.where(my2DArray > value) or indices_true = numpy.where(my2DArray == value2)	Returns a tuple with arrays of the indices where the condition (> value, == value2, etc.) is true. If you want the array of row indices, you would need to do indices_true[0]. Likewise, for the columns, indices_true[1]. Will return empty arrays within the tuple if no elements meet the condition. Will also return a tuple, even for 1D arrays.