javicid

This is to avoid the following warning: ``` Tests KFold.split() from single subset, shuffle=False and uneven ... /dislib/tests/test_kfold.py:67: DeprecationWarning: Please use assertEqual instead. self.assertEquals(len_x_train, len_y_train) ```

Block size of irregular arrays changes after some indexing operations. It would be better to keep the block size of the original array when possible.

C-SVM predict not deterministic

1

In rare cases, C-SVM predict returns different labels for the same data. This happens especially in CSVMTest.test_sparse, which compares the labels returned by predict when using sparse and dense data...

Create a test that checks that RandomForest produces the same results regardless of the data structures employed to store data (dense or sparse).

Add convergence argument to K-means such that it can be executed without checking for convergence. KMeans.centers should then be kept as future objects until the user access them.

Subclass NumPy arrays

1

ds-arrays could inherit from NumPy arrays. This would allow to automatically parallelize any existing algorithm or library that uses NumPy arrays internally. However, we should discuss the feasibility of this...

The current implementation of SVD has the following limitations: - Convergence checking needs to synchronize data (could be improved with exceptions mechanism) - Convergence criteria should be revised because it...

Standarize nomenclature in the source code and documentation according to: 1. `dislib.Array` instances are referred to as `ds-array`. 2. NumPy array instances are referred to as `ndarray`. 3. `x` and...

Modify array indexing routines to raise not implemented exceptions in the not supported cases.