Document

crosstab(*args, levels=None, sparse=False)

When len(args) > 1, the array computed by this function is often referred to as a contingency table .

The arguments must be sequences with the same length. The second return value, count, is an integer array with len(args) dimensions. If levels is None, the shape of count is (n0, n1, ...), where nk is the number of unique elements in args[k].

Notes

Parameters

*args : sequences: A sequence of sequences whose unique aligned elements are to be counted. The sequences in args must all be the same length.
levels : sequence, optional: If levels is given, it must be a sequence that is the same length as args. Each element in levels is either a sequence or None. If it is a sequence, it gives the values in the corresponding sequence in args that are to be counted. If any value in the sequences in args does not occur in the corresponding sequence in levels, that value is ignored and not counted in the returned array count. The default value of levels for args[i] is np.unique(args[i])
sparse : bool, optional: If True, return a sparse matrix. The matrix will be an instance of the scipy.sparse.coo_matrix class. Because SciPy's sparse matrices must be 2-d, only two input sequences are allowed when sparse is True. Default is False.

Returns

res : CrosstabResult

An object containing the following attributes:

elements: elements
count: count

Return table of counts for each possible unique combination in *args.

Examples

from scipy.stats.contingency import crosstab

Given the lists `a` and `x`, create a contingency table that counts the frequencies of the corresponding pairs.

a = ['A', 'B', 'A', 'A', 'B', 'B', 'A', 'A', 'B', 'B']
x = ['X', 'X', 'X', 'Y', 'Z', 'Z', 'Y', 'Y', 'Z', 'Z']
res = crosstab(a, x)
avals, xvals = res.elements
avals

xvals

res.count

So `('A', 'X')` occurs twice, `('A', 'Y')` occurs three times, etc.

Higher dimensional contingency tables can be created.

p = [0, 0, 0, 0, 1, 1, 1, 0, 0, 1]
res = crosstab(a, x, p)
res.count

res.count.shape

The values to be counted can be set by using the `levels` argument. It allows the elements of interest in each input sequence to be given explicitly instead finding the unique elements of the sequence.

For example, suppose one of the arguments is an array containing the answers to a survey question, with integer values 1 to 4. Even if the value 1 does not occur in the data, we want an entry for it in the table.

q1 = [2, 3, 3, 2, 4, 4, 2, 3, 4, 4, 4, 3, 3, 3, 4]  # 1 does not occur.
q2 = [4, 4, 2, 2, 2, 4, 1, 1, 2, 2, 4, 2, 2, 2, 4]  # 3 does not occur.
options = [1, 2, 3, 4]
res = crosstab(q1, q2, levels=(options, options))
res.count

If `levels` is given, but an element of `levels` is None, the unique values of the corresponding argument are used. For example,

res = crosstab(q1, q2, levels=(None, options))
res.elements

res.count

If we want to ignore the pairs where 4 occurs in ``q2``, we can give just the values [1, 2] to `levels`, and the 4 will be ignored:

res = crosstab(q1, q2, levels=(None, [1, 2]))
res.elements

res.count

Finally, let's repeat the first example, but return a sparse matrix:

res = crosstab(a, x, sparse=True)
res.count

res.count.A

See :

Local connectivity graph

Hover to see nodes names; edges to Self not shown, Caped at 50 nodes.

Using a canvas is more power efficient and can get hundred of nodes ; but does not allow hyperlinks; , arrows or text (beyond on hover)

SVG is more flexible but power hungry; and does not scale well to 50 + nodes.

All aboves nodes referred to, (or are referred from) current nodes; Edges from Self to other have been omitted (or all nodes would be connected to the central node "self" which is not useful). Nodes are colored by the library they belong to, and scaled with the number of references pointing them

GitHub : /scipy/stats/_crosstab.py#10
type: <class 'function'>
Commit:

Notes

Parameters

Returns

See Also

Examples

Local connectivity graph