`skcriteria.core.data` module¶

Data abstraction layer.

This module defines the DecisionMatrix object, which internally encompasses the alternative matrix, weights and objectives (MIN, MAX) of the criteria.

class skcriteria.core.data.DecisionMatrix(data_df, objectives, weights)[source]

Representation of all data needed in the MCDA analysis.

This object gathers everything necessary to represent a data set used in MCDA:

• An alternative matrix where each row is an alternative and each column is of a different criteria.

• An optimization objective (Minimize, Maximize) for each criterion.

• A weight for each criterion.

• An independent type of data for each criterion

DecisionMatrix has two main forms of construction:

1. Use the default constructor of the DecisionMatrix class `pandas.DataFrame` where the index is the alternatives and the columns are the criteria; an iterable with the objectives with the same amount of elements that columns/criteria has the dataframe; and an iterable with the weights also with the same amount of elements as criteria.

```
```
```>>> import pandas as pd
>>> from skcriteria import DecisionMatrix, mkdm
```
```>>> data_df = pd.DataFrame(
...     [[1, 2, 3], [4, 5, 6]],
...     index=["A0", "A1"],
...     columns=["C0", "C1", "C2"]
... )
>>> objectives = [min, max, min]
>>> weights = [1, 1, 1]
```
```>>> dm = DecisionMatrix(data_df, objectives, weights)
>>> dm
C0[▼ 1.0] C1[▲ 1.0] C2[▲ 1.0]
A0         1         2         3
A1         4         5         6
[2 Alternatives x 3 Criteria]
```
1. Use the classmethod DecisionMatrix.from_mcda_data which requests the data in a more natural way for this type of analysis (the weights, the criteria / alternative names, and the data types are optional)

```>>> DecisionMatrix.from_mcda_data(
...     [[1, 2, 3], [4, 5, 6]],
...     [min, max, min],
...     [1, 1, 1])
C0[▼ 1.0] C1[▲ 1.0] C2[▲ 1.0]
A0         1         2         3
A1         4         5         6
[2 Alternatives x 3 Criteria]
```

For simplicity a function is offered at the module level analogous to `from_mcda_data` called `mkdm` (make decision matrix).

Parameters:
• data_df (`pandas.DatFrame`) – Dataframe where the index is the alternatives and the columns are the criteria.

• objectives (`numpy.ndarray`) – Aan iterable with the targets with sense of optimality of every criteria (You can use any alias defined in Objective) the same length as columns/criteria has the data_df.

• weights (`numpy.ndarray`) – An iterable with the weights also with the same amount of elements as criteria.

classmethod from_mcda_data(matrix, objectives, *, weights=None, alternatives=None, criteria=None, dtypes=None)[source]

Create a new DecisionMatrix object.

This method receives the parts of the matrix, in what conceptually the matrix of alternatives is usually divided

Parameters:
• matrix (Iterable) – The matrix of alternatives. Where every row is an alternative and every column is a criteria.

• objectives (Iterable) – The array with the sense of optimality of every criteria. You can use any alias provided by the objective class.

• weights (Iterable o None (default `None`)) – Optional weights of the criteria. If is `None` all the criteria are weighted with 1.

• alternatives (Iterable o None (default `None`)) – Optional names of the alternatives. If is `None`, al the alternatives are names “A[n]” where n is the number of the row of matrix statring at 0.

• criteria (Iterable o None (default `None`)) – Optional names of the criteria. If is `None`, al the alternatives are names “C[m]” where m is the number of the columns of matrix statring at 0.

• dtypes (Iterable o None (default `None`)) – Optional types of the criteria. If is None, the type is inferred automatically by pandas.

Returns:

A new decision matrix.

Return type:

`DecisionMatrix`

Example

```>>> DecisionMatrix.from_mcda_data(
...     [[1, 2, 3], [4, 5, 6]],
...     [min, max, min],
...     [1, 1, 1])
C0[▼ 1.0] C1[▲ 1.0] C2[▲ 1.0]
A0         1         2         3
A1         4         5         6
[2 Alternatives x 3 Criteria]
```

For simplicity a function is offered at the module level analogous to `from_mcda_data` called `mkdm` (make decision matrix).

Notes

This functionality generates more sensitive defaults than using the constructor of the DecisionMatrix class but is slower.

property alternatives

Names of the alternatives.

From this array you can also access the values of the alternatives as `pandas.Series`.

property criteria

Names of the criteria.

From this array you can also access the values of the criteria as `pandas.Series`.

property weights

Weights of the criteria.

property objectives

Objectives of the criteria as `Objective` instances.

property minwhere

Mask with value True if the criterion is to be minimized.

property maxwhere

Mask with value True if the criterion is to be maximized.

property iobjectives

Objectives of the criteria as `int`.

• Minimize = Objective.MIN.value

• Maximize = Objective.MAX.value

property matrix

Alternatives matrix as pandas DataFrame.

The matrix excludes weights and objectives.

If you want to create a DataFrame with objectives and weights, use `DecisionMatrix.to_dataframe()`

property dtypes

Dtypes of the criteria.

property plot

Plot accessor.

property stats

Descriptive statistics accessor.

property dominance

Dominance information accessor.

copy(**kwargs)[source]

Return a deep copy of the current DecisionMatrix.

This method is also useful for manually modifying the values of the DecisionMatrix object.

Parameters:

kwargs – The same parameters supported by `from_mcda_data()`. The values provided replace the existing ones in the object to be copied.

Returns:

A new decision matrix.

Return type:

`DecisionMatrix`

to_dataframe()[source]

Convert the entire DecisionMatrix into a dataframe.

The objectives and weights ara added as rows before the alternatives.

Returns:

A Decision matrix as pandas DataFrame.

Return type:

`pd.DataFrame`

Example

```>>> dm = DecisionMatrix.from_mcda_data(
>>> dm
...     [[1, 2, 3], [4, 5, 6]],
...     [min, max, min],
...     [1, 1, 1])
C0[▼ 1.0] C1[▲ 1.0] C2[▲ 1.0]
A0         1         2         3
A1         4         5         6

>>> dm.to_dataframe()
C0   C1   C2
objectives  MIN  MAX  MIN
weights     1.0  1.0  1.0
A0            1    2    3
A1            4    5    6
```
to_dict()[source]

Return a dict representation of the data.

All the values are represented as numpy array.

describe(**kwargs)[source]

Generate descriptive statistics.

Descriptive statistics include those that summarize the central tendency, dispersion and shape of a dataset’s distribution, excluding `NaN` values.

Parameters:

pandas.DataFrame.describe(). (Same parameters as) –

Returns:

Summary statistics of DecisionMatrix provided.

Return type:

`pandas.DataFrame`

property shape

Return a tuple with (number_of_alternatives, number_of_criteria).

dm.shape <==> np.shape(dm)

diff(other, rtol=1e-05, atol=1e-08, equal_nan=True, check_dtypes=False)[source]

Return the difference between two objects within a tolerance.

This method should be implemented by subclasses to define how differences between objects are calculated.

The tolerance parameters rtol and atol, equal_nan, and check_dtypes are provided to be used by the numpy and pandas equality functions. These parameters allow you to customize the behavior of the equality comparison, such as setting the relative and absolute tolerance for numeric comparisons, considering NaN values as equal, and checking for the data type of the objects being compared.

Parameters:
• other (object) – The object to compare to.

• rtol (float, optional) – The relative tolerance parameter. Default is 1e-05.

• atol (float, optional) – The absolute tolerance parameter. Default is 1e-08.

• equal_nan (bool, optional) – Whether to consider NaN values as equal. Default is True.

• check_dtypes (bool, optional) – Whether to check the data type of the objects. Default is False.

Returns:

The difference between the current and the other object.

Return type:

the_diff

`equals`, `aequals`, `numpy.isclose()`, `numpy.all()`, `numpy.any()`, `numpy.equal()`, `numpy.allclose()`

Notes

The tolerance values are positive, typically very small numbers. The relative difference (rtol * abs(b)) and the absolute difference atol are added together to compare against the absolute difference between a and b.

NaNs are treated as equal if they are in the same place and if `equal_nan=True`. Infs are treated as equal if they are in the same place and of the same sign in both arrays.

property loc

Access a group of alternatives and criteria by label(s) or a boolean array.

`.loc[]` is primarily alternative label based, but may also be used with a boolean array.

Unlike DataFrames, ìloc` of `DecisionMatrix` always returns an instance of `DecisionMatrix`.

property iloc

Purely integer-location based indexing for selection by position.

`.iloc[]` is primarily integer position based (from `0` to `length-1` of the axis), but may also be used with a boolean array.

Unlike DataFrames, ìloc` of `DecisionMatrix` always returns an instance of `DecisionMatrix`.

skcriteria.core.data.mkdm(matrix, objectives, *, weights=None, alternatives=None, criteria=None, dtypes=None)[source]

Create a new DecisionMatrix object.

This method receives the parts of the matrix, in what conceptually the matrix of alternatives is usually divided

Parameters:
• matrix (Iterable) – The matrix of alternatives. Where every row is an alternative and every column is a criteria.

• objectives (Iterable) – The array with the sense of optimality of every criteria. You can use any alias provided by the objective class.

• weights (Iterable o None (default `None`)) – Optional weights of the criteria. If is `None` all the criteria are weighted with 1.

• alternatives (Iterable o None (default `None`)) – Optional names of the alternatives. If is `None`, al the alternatives are names “A[n]” where n is the number of the row of matrix statring at 0.

• criteria (Iterable o None (default `None`)) – Optional names of the criteria. If is `None`, al the alternatives are names “C[m]” where m is the number of the columns of matrix statring at 0.

• dtypes (Iterable o None (default `None`)) – Optional types of the criteria. If is None, the type is inferred automatically by pandas.

Returns:

A new decision matrix.

Return type:

`DecisionMatrix`

Example

```>>> DecisionMatrix.from_mcda_data(
...     [[1, 2, 3], [4, 5, 6]],
...     [min, max, min],
...     [1, 1, 1])
C0[▼ 1.0] C1[▲ 1.0] C2[▲ 1.0]
A0         1         2         3
A1         4         5         6
[2 Alternatives x 3 Criteria]
```

For simplicity a function is offered at the module level analogous to `from_mcda_data` called `mkdm` (make decision matrix).

Notes

This functionality generates more sensitive defaults than using the constructor of the DecisionMatrix class but is slower.