Add inverse boxcox transform functionality to Box Cox Transformer

[SangamSwadiK]

Reference Issues/PRs:
Issue : #449 (Add inverse Boxcox transform functionality)

What does this implement/fix?
Adds an inverse transform method to Boxcox transform

Please let me know if I missed something / did it incorrectly, ill fix it asap.

[solegalli]

Hi @SangamSwadiK

The code looks great. Thank you!

We would have to expand this test to ensure that the inverse_transform returns the desired functionality:

feature_engine/tests/test_transformation/test_boxcox_transformer.py

Lines 8 to 24 in c16e989

	def test_automatically_finds_variables(df_vartypes):
	# test case 1: automatically select variables
	transformer = BoxCoxTransformer(variables=None)
	X = transformer.fit_transform(df_vartypes)
	# expected output
	transf_df = df_vartypes.copy()
	transf_df["Age"] = [9.78731, 10.1666, 9.40189, 9.0099]
	transf_df["Marks"] = [-0.101687, -0.207092, -0.316843, -0.431788]
	# test init params
	assert transformer.variables is None
	# test fit attr
	assert transformer.variables_ == ["Age", "Marks"]
	assert transformer.n_features_in_ == 5
	# test transform output
	pd.testing.assert_frame_equal(X, transf_df)

Would you be able to do that?

For inspiration you could look at how we do it for the LogTransformer here:

feature_engine/tests/test_transformation/test_log_transformer.py

Lines 8 to 35 in c16e989

	def test_log_base_e_plus_automatically_find_variables(df_vartypes):
	# test case 1: log base e, automatically select variables
	transformer = LogTransformer(base="e", variables=None)
	X = transformer.fit_transform(df_vartypes)
	# expected output
	transf_df = df_vartypes.copy()
	transf_df["Age"] = [2.99573, 3.04452, 2.94444, 2.89037]
	transf_df["Marks"] = [-0.105361, -0.223144, -0.356675, -0.510826]
	# test init params
	assert transformer.base == "e"
	assert transformer.variables is None
	# test fit attr
	assert transformer.variables_ == ["Age", "Marks"]
	assert transformer.n_features_in_ == 5
	# test transform output
	pd.testing.assert_frame_equal(X, transf_df)
	# test inverse_transform
	Xit = transformer.inverse_transform(X)
	# convert numbers to original format.
	Xit["Age"] = Xit["Age"].round().astype("int64")
	Xit["Marks"] = Xit["Marks"].round(1)
	# test
	pd.testing.assert_frame_equal(Xit, df_vartypes)

[SangamSwadiK]

Hi @SangamSwadiK

The code looks great. Thank you!

We would have to expand this test to ensure that the inverse_transform returns the desired functionality:

feature_engine/tests/test_transformation/test_boxcox_transformer.py

Lines 8 to 24 in c16e989

	def test_automatically_finds_variables(df_vartypes):
	# test case 1: automatically select variables
	transformer = BoxCoxTransformer(variables=None)
	X = transformer.fit_transform(df_vartypes)
	# expected output
	transf_df = df_vartypes.copy()
	transf_df["Age"] = [9.78731, 10.1666, 9.40189, 9.0099]
	transf_df["Marks"] = [-0.101687, -0.207092, -0.316843, -0.431788]
	# test init params
	assert transformer.variables is None
	# test fit attr
	assert transformer.variables_ == ["Age", "Marks"]
	assert transformer.n_features_in_ == 5
	# test transform output
	pd.testing.assert_frame_equal(X, transf_df)

Would you be able to do that?

For inspiration you could look at how we do it for the LogTransformer here:

feature_engine/tests/test_transformation/test_log_transformer.py

Lines 8 to 35 in c16e989

	def test_log_base_e_plus_automatically_find_variables(df_vartypes):
	# test case 1: log base e, automatically select variables
	transformer = LogTransformer(base="e", variables=None)
	X = transformer.fit_transform(df_vartypes)
	# expected output
	transf_df = df_vartypes.copy()
	transf_df["Age"] = [2.99573, 3.04452, 2.94444, 2.89037]
	transf_df["Marks"] = [-0.105361, -0.223144, -0.356675, -0.510826]
	# test init params
	assert transformer.base == "e"
	assert transformer.variables is None
	# test fit attr
	assert transformer.variables_ == ["Age", "Marks"]
	assert transformer.n_features_in_ == 5
	# test transform output
	pd.testing.assert_frame_equal(X, transf_df)
	# test inverse_transform
	Xit = transformer.inverse_transform(X)
	# convert numbers to original format.
	Xit["Age"] = Xit["Age"].round().astype("int64")
	Xit["Marks"] = Xit["Marks"].round(1)
	# test
	pd.testing.assert_frame_equal(Xit, df_vartypes)

@solegalli I've added test for inverse transform. Please let me know if changes are required. I'll do it.
Thanks.

[solegalli]

Hi @SangamSwadiK

thanks a lot for the quick turnaround!

This is almost ready. I should have mentioned before that we need to add the inverse_transform to the docstings of the class.

See for example how we bring the documentation forward for the log transformer using {inverse_transform} here:

feature_engine/feature_engine/transformation/log.py

Lines 64 to 73 in c16e989

	Methods
	-------
	{fit}
	{fit_transform}
	{inverse_transform}
	transform:
	Transform the variables using the logarithm.

which we import from our base docstrings like we do here:

feature_engine/feature_engine/transformation/log.py

Lines 10 to 14 in c16e989

	from feature_engine._docstrings.methods import (
	_fit_not_learn_docstring,
	_fit_transform_docstring,
	_inverse_transform_docstring,
	)

So if you could add the 2 lines of code above and also rephrase the sentence in the inverse_transform method that I point to in a following message, then, this is ready to go!!

Thank you so much!

[solegalli]

Here, for convention we write the following:

Convert the data back to the original representation.

Instead of what you wrote, that makes total sense. It's just that we want to follow scikit-learn conventions.

[SangamSwadiK]

Hi @SangamSwadiK

thanks a lot for the quick turnaround!

This is almost ready. I should have mentioned before that we need to add the inverse_transform to the docstings of the class.

See for example how we bring the documentation forward for the log transformer using {inverse_transform} here:

feature_engine/feature_engine/transformation/log.py

Lines 64 to 73 in c16e989

	Methods
	-------
	{fit}
	{fit_transform}
	{inverse_transform}
	transform:
	Transform the variables using the logarithm.

which we import from our base docstrings like we do here:

feature_engine/feature_engine/transformation/log.py

Lines 10 to 14 in c16e989

	from feature_engine._docstrings.methods import (
	_fit_not_learn_docstring,
	_fit_transform_docstring,
	_inverse_transform_docstring,
	)

So if you could add the 2 lines of code above and also rephrase the sentence in the inverse_transform method that I point to in a following message, then, this is ready to go!!

Thank you so much!

@solegalli Sorry, I forgot to update the class docs, I've did the changes now.
Thanks.

[solegalli]

Thank you @SangamSwadiK !

Merging as we speak :)