v0.2.4 (May 26, 2023)#

These are minor changes in v0.2.3 that shows fugs fixed, features and improvements.

  • API change Change performed in watex.geology.core. The class Base is renamed as watex.geology.core.GeoBase.

  • Fix Bug fixed in watex.utils.plot_logging() when imputing NaN values in the data. Indeed, henceforth all the column of the data that is composed entirely with NaN is removed automatically. This fixes the bug occurs in when the scaled feature dimension are higher than the original ones.

  • Fix Bug fixed in watex.utils.to_numeric_dtypes() when sanitizing the dataframe with integer columns. In such case, the data sanitizing is not possible. It henceforth passes rather than raising error. However when column are mixed ( numeric and string ), all should be converted to string before sanitization.

  • Feature watex.geology.drilling.DSBorehole works with data collected in a single borehole. For instance, the arrangement of h502 data in watex.datasets.load_hlogs(). Data columns are the all attributes of the object and any non-alphateic character is replaced by _. For instance, a column name layer thickness should have an attribute named layer_thickness.

  • Feature watex.geology.drilling.DSBoreholes works similarly as watex.geology.drilling.DSBorehole but uses multiple borehole data. note the 's' at the end of the class. Here, each borehole (row) is an object which encompasses all columns as attributes. To have full control of how data must be retrieved as an object, holeid parameter must be set. For instance, to retrieve the borehole with ID equals to bx02, after fitting the class with appropriate parameters, attibute hole depth ( if exist in the data) can be retrieved as self.hole.bx02.hole_depth.

  • Feature Add watex.utils.funcutils.replace_data() to replace items in the data n- times. This is useful for data duplication

  • Feature add watex.utils.funcutils.make_obj_consistent_if. Function combine default values to item to create default consistent iterable objects. This is valid if the size of item does not fit the number of expected iterable objects. Commonly this work in synergy with watex.utils.funcutils.random_sampling().

  • Feature watex.utils.funcutils.convert_value_in() converts value based on the reference unit. Mostly used the known SI units like meters (m) `, `grammes (g) etc.

  • Enhancement Supplement boreholes data are added to watex.datasets.load_hlogs(). They include boreholes 'h805' ,'h1102' ,'h1405', 'h2602','h803', 'h1104', 'h604', 'h2003' and 'h1602'.

  • Enhancement Function watex.utils.to_numeric_dtypes() accepts drop_nan_columns passed as True by default with dropped the NaN rows and NaN columns in the data. Moreover, reset_index and drop_index are available to new data index management.

  • Enhancement Function watex.datasets.load_mxs() enforces attribute X and y when object is retrieved. For instance, fetching the the training set scaled X should be:

    >>> import watex as wx
>>> X_scaled = wx.datasets.load_mxs (key ='scale').X
>>> # or by merely calling the boilerplate `fetch_data` function as
>>> X_scaled =wx.fetch_data ('mxs', key ='scale').X

    It also accepts test_ratio to split the training and test sets via split_X_y=True. Note that return_X_y=True is not necessary when split_X_y is set to True.

  • Enhancement watex.transformers.featurize_X accepts model as parameter. It is useful to featurize the test data separately from the train data to avoid leakage information. If sparse parameter is set to True, the training and testing data are outputted in sparse coordinates format or compressed sparse row format that is controlled by the sparsity parameter.