Imbalanced Learning Regression

https://imbalancedlearningregression.readthedocs.io/en/latest/
Admin
Imbalanced Learning Regression is a Python library that implements sampling techniques for...
Tokens:14,137
Snippets:75
Trust Score:9.6
Update:1 month ago
Show doc for...
Tokens:
Raw
### Install developer version from GitHub

Source: https://imbalancedlearningregression.readthedocs.io/en/latest/intro.html

Install the latest developer version directly from the GitHub repository.

```bash
$ pip install git+https://github.com/paobranco/ImbalancedLearningRegression.git

```

--------------------------------

### Install pypi release

Source: https://imbalancedlearningregression.readthedocs.io/en/latest/intro.html

Install the package from PyPI using pip.

```bash
$ pip install ImbalancedLearningRegression

```

--------------------------------

### Python example: Using cnn() to under-sample a dataset

Source: https://imbalancedlearningregression.readthedocs.io/en/latest/cnn.html

Shows how to import the cnn function, load a dataset from a URL, and apply under-sampling with default parameters. The example uses the housing dataset and specifies the target column 'SalePrice'.

```python
>>> from ImbalancedLearningRegression import cnn
>>> housing = pandas.read_csv("https://raw.githubusercontent.com/paobranco/ImbalancedLearningRegression/master/data/housing.csv")
>>> housing_cnn = cnn(data = housing, y = "SalePrice")
```

--------------------------------

### Example: Using ro() to over-sample the housing dataset

Source: https://imbalancedlearningregression.readthedocs.io/en/latest/ro.html

Shows how to import the ro function, load a dataset from a URL, and apply random over-sampling to the target column 'SalePrice'.

```python
>>> from ImbalancedLearningRegression import ro
>>> housing = pandas.read_csv("https://raw.githubusercontent.com/paobranco/ImbalancedLearningRegression/master/data/housing.csv")
>>> housing_ro = ro(data = housing, y = "SalePrice")

```

--------------------------------

### Perform ADASYN over-sampling

Source: https://imbalancedlearningregression.readthedocs.io/en/latest/adasyn.html

Example showing how to import the adasyn function and apply it to a housing dataset to re-sample based on the 'SalePrice' target variable.

```python
>>> from ImbalancedLearningRegression import adasyn
>>> housing = pandas.read_csv("https://raw.githubusercontent.com/paobranco/ImbalancedLearningRegression/master/data/housing.csv")
>>> housing_adasyn = adasyn(data = housing, y = "SalePrice")
```

--------------------------------

### tomeklinks() usage example

Source: https://imbalancedlearningregression.readthedocs.io/en/latest/tomeklinks.html

Example showing how to import and use the tomeklinks() function from ImbalancedLearningRegression. It loads the housing dataset from a URL and applies TomekLinks under-sampling with default parameters.

```python
>>> from ImbalancedLearningRegression import tomeklinks
>>> housing = pandas.read_csv("https://raw.githubusercontent.com/paobranco/ImbalancedLearningRegression/master/data/housing.csv")
>>> housing_tomeklinks = tomeklinks(data = housing, y = "SalePrice")

```

--------------------------------

### random_under() usage example

Source: https://imbalancedlearningregression.readthedocs.io/en/latest/ru.html

Imports the random_under function, loads the housing dataset from a URL, and applies random under-sampling with default parameters.

```python
>>> from ImbalancedLearningRegression import random_under
>>> housing = pandas.read_csv("https://raw.githubusercontent.com/paobranco/ImbalancedLearningRegression/master/data/housing.csv")
>>> housing_ru = random_under(data = housing, y = "SalePrice")

```

--------------------------------

### Usage example with Random Over-sampling and Gaussian Noise

Source: https://imbalancedlearningregression.readthedocs.io/en/latest/intro.html

Demonstrates loading the housing dataset and applying random over-sampling (ro) and Gaussian noise (gn) techniques using the ImbalancedLearningRegression package.

```python
>>> ## load libraries
>>> import pandas
>>> import ImbalancedLearningRegression as iblr

>>> ## load data
>>> housing = pandas.read_csv("https://raw.githubusercontent.com/paobranco/ImbalancedLearningRegression/master/data/housing.csv")

>>> ## conduct Random Over-sampling
>>> housing_ro = iblr.ro(data = housing, y = "SalePrice")

>>> ## conduct Introduction of Gaussian Noise
>>> housing_gn = iblr.gn(data = housing, y = "SalePrice")

```

--------------------------------

### enn() usage example with housing dataset

Source: https://imbalancedlearningregression.readthedocs.io/en/latest/enn.html

Imports the enn function, loads the housing dataset from a URL, and applies ENN under-sampling to the 'SalePrice' target column. The result is stored in housing_enn.

```python
>>> from ImbalancedLearningRegression import enn
>>> housing = pandas.read_csv("https://raw.githubusercontent.com/paobranco/ImbalancedLearningRegression/master/data/housing.csv")
>>> housing_enn = enn(data = housing, y = "SalePrice")
```

--------------------------------

### Import and apply gn() to a housing dataset

Source: https://imbalancedlearningregression.readthedocs.io/en/latest/gn.html

Shows how to import the gn function from ImbalancedLearningRegression, load a housing dataset from a URL, and apply gn() with default parameters to re-sample the data. The result is stored in housing_gn.

```python
>>> from ImbalancedLearningRegression import gn
>>> housing = pandas.read_csv("https://raw.githubusercontent.com/paobranco/ImbalancedLearningRegression/master/data/housing.csv")
>>> housing_gn = gn(data = housing, y = "SalePrice")

```

--------------------------------

### random_under(data, y, samp_method='balance', drop_na_col=True, drop_na_row=True, replacement=False, manual_perc=False, perc_u=-1, rel_thres=0.5, rel_method='auto', rel_xtrm_type='both', rel_coef=1.5, rel_ctrl_pts_rg=None)

Source: https://imbalancedlearningregression.readthedocs.io/en/latest/ru.html

Randomly selects a subset of majority samples to under-sample the dataset. Returns a re-sampled Pandas dataframe.

```APIDOC
## random_under(data, y, samp_method='balance', drop_na_col=True, drop_na_row=True, replacement=False, manual_perc=False, perc_u=-1, rel_thres=0.5, rel_method='auto', rel_xtrm_type='both', rel_coef=1.5, rel_ctrl_pts_rg=None)

### Description
Randomly selects a subset of majority samples to under-sample the dataset. Returns a re-sampled Pandas dataframe.

### Method
Function call

### Endpoint
random_under(data, y, samp_method='balance', drop_na_col=True, drop_na_row=True, replacement=False, manual_perc=False, perc_u=-1, rel_thres=0.5, rel_method='auto', rel_xtrm_type='both', rel_coef=1.5, rel_ctrl_pts_rg=None)

### Parameters
- **data** (Pandas dataframe) - Required - The dataset to re-sample.
- **y** (str) - Required - Column name of the target variable in the Pandas dataframe.
- **samp_method** (str) - Optional - Method to determine re-sampling percentage. Either `balance` or `extreme`.
- **drop_na_col** (bool) - Optional - Determine whether or not automatically drop columns containing NaN values. The data frame should not contain any missing values, so it is suggested to keep it as default.
- **drop_na_row** (bool) - Optional - Determine whether or not automatically drop rows containing NaN values. The data frame should not contain any missing values, so it is suggested to keep it as default.
- **replacement** (bool) - Optional - Randomly select sample to duplicate: with or without replacement.
- **manual_perc** (bool) - Optional - Keep the same percentage of re-sampling for all bins. If `True`, `perc_u` is required to be a real number between 0 and 1 (0, 1).
- **perc_u** (float) - Optional - User-specified fixed percentage of under-sampling for all bins. Must be a real number between 0 and 1 (0, 1) if `manual_perc = True`.
- **rel_thres** (float) - Optional - Relevance threshold, above which a sample is considered rare. Must be a real number between 0 and 1 (0, 1].
- **rel_method** (str) - Optional - Method to define the relevance function, either `auto` or `manual`. If `manual`, must specify `rel_ctrl_pts_rg`.
- **rel_xtrm_type** (str) - Optional - Distribution focus, `high`, `low`, or `both`. If `high`, rare cases having small y values will be considerd as normal, and vise versa.
- **rel_coef** (float) - Optional - Coefficient for box plot.
- **rel_ctrl_pts_rg** (2D array) - Optional - Manually specify the regions of interest. See SMOGN advanced example for more details.

### Returns
Re-sampled dataset (Pandas dataframe).

### Raises
- **ValueError** - If an input attribute has wrong data type or invalid value, or relevance values are all zero or all one, or under_sampled data contains missing values.

### Example
python
>>> from ImbalancedLearningRegression import random_under
>>> housing = pandas.read_csv("https://raw.githubusercontent.com/paobranco/ImbalancedLearningRegression/master/data/housing.csv")
>>> housing_ru = random_under(data = housing, y = "SalePrice")

```

--------------------------------

### Apply smote() to housing dataset

Source: https://imbalancedlearningregression.readthedocs.io/en/latest/smote.html

Basic usage of the smote() function from the ImbalancedLearningRegression package. Loads the housing dataset from a URL and applies SMOTE with default parameters (k=5, samp_method='balance'). The result is stored in housing_smote.

```python
>>> from ImbalancedLearningRegression import smote
>>> housing = pandas.read_csv("https://raw.githubusercontent.com/paobranco/ImbalancedLearningRegression/master/data/housing.csv")
>>> housing_smote = smote(data = housing, y = "SalePrice")
```

--------------------------------

### cnn()

Source: https://imbalancedlearningregression.readthedocs.io/en/latest/cnn.html

Condensed Nearest Neighbor under-sampling method. It condenses the majority set by selecting a subset of majority samples from the original majority set.

```APIDOC
## cnn()

### Description
Condensed Nearest Neighbor is an under-sampling method that condenses the majority set by selecting a subset of majority samples from the original majority set.

### Method
Function call

### Signature
cnn(data, y, samp_method='balance', drop_na_col=True, drop_na_row=True, n_seed=1, rel_thres=0.5, rel_method='auto', rel_xtrm_type='both', rel_coef=1.5, rel_ctrl_pts_rg=None, k=1, n_jobs=1, k_neighbors_classifier=None)

### Parameters
- **data** (Pandas dataframe) - Required - Pandas dataframe, the dataset to re-sample.
- **y** (str) - Required - Column name of the target variable in the Pandas dataframe.
- **samp_method** (str) - Optional - Method to determine re-sampling percentage. Either `balance` or `extreme`.
- **drop_na_col** (bool) - Optional - Determine whether or not automatically drop columns containing NaN values. The data frame should not contain any missing values, so it is suggested to keep it as default.
- **drop_na_row** (bool) - Optional - Determine whether or not automatically drop rows containing NaN values. The data frame should not contain any missing values, so it is suggested to keep it as default.
- **n_seed** (int) - Optional - Number of majority samples put into STORE at the beginning of under-sampling each normal bin. Must be a positive integer.
- **rel_thres** (float) - Optional - Relevance threshold, above which a sample is considered rare. Must be a real number between 0 and 1 (0, 1].
- **rel_method** (str) - Optional - Method to define the relevance function, either `auto` or `manual`. If `manual`, must specify `rel_ctrl_pts_rg`.
- **rel_xtrm_type** (str) - Optional - Distribution focus, `high`, `low`, or `both`. If `high`, rare cases having small y values will be considerd as normal, and vise versa.
- **rel_coef** (float) - Optional - Coefficient for box plot.
- **rel_ctrl_pts_rg** (2D array) - Optional - Manually specify the regions of interest. See SMOGN advanced example for more details.
- **k** (int) - Optional - The number of neighbors considered. Must be a positive integer.
- **n_jobs** (int) - Optional - The number of parallel jobs to run for neighbors search. Must be an integer. See sklearn.neighbors.KNeighborsClassifier for more details.
- **k_neighbors_classifier** (KNeighborsClassifier) - Optional - If users want to define more parameters of KNeighborsClassifier, such as `weights`, `algorithm`, `leaf_size`, and `metric`, they can create an instance of KNeighborsClassifier and pass it to this method. In that case, setting `k` and `n_jobs` will have no effect.

### Returns
Re-sampled dataset. Type: Pandas dataframe.

### Raises
- **ValueError** - If an input attribute has wrong data type or invalid value, or relevance values are all zero or all one, or once the index of a sample exists in both STORE and GRABBAG.

### Example
```python
>>> from ImbalancedLearningRegression import cnn
>>> housing = pandas.read_csv("https://raw.githubusercontent.com/paobranco/ImbalancedLearningRegression/master/data/housing.csv")
>>> housing_cnn = cnn(data = housing, y = "SalePrice")
```
```

--------------------------------

### smote(data, y, k=5, samp_method='balance', drop_na_col=True, drop_na_row=True, rel_thres=0.5, rel_method='auto', rel_xtrm_type='both', rel_coef=1.5, rel_ctrl_pts_rg=None)

Source: https://imbalancedlearningregression.readthedocs.io/en/latest/smote.html

The smote() function applies the Synthetic Minority Oversampling Technique to a dataset for regression. It takes a pandas dataframe and the target column name, and returns a re-sampled dataframe.

```APIDOC
## smote(data, y, k=5, samp_method='balance', drop_na_col=True, drop_na_row=True, rel_thres=0.5, rel_method='auto', rel_xtrm_type='both', rel_coef=1.5, rel_ctrl_pts_rg=None)

### Description
Synthetic Minority Oversampling Technique. SMOTE is an over-sampling method that synthesizes new samples by using one of the neighbors of a seed sample.

### Parameters
- **data** (Pandas dataframe) - Required - The dataset to re-sample.
- **y** (str) - Required - Column name of the target variable in the Pandas dataframe.
- **k** (int) - Optional - The number of neighbors considered. Must be a positive integer. Default: 5.
- **samp_method** (str) - Optional - Method to determine re-sampling percentage. Either `balance` or `extreme`. Default: 'balance'.
- **drop_na_col** (bool) - Optional - Determine whether or not automatically drop columns containing NaN values. The data frame should not contain any missing values, so it is suggested to keep it as default. Default: True.
- **drop_na_row** (bool) - Optional - Determine whether or not automatically drop rows containing NaN values. The data frame should not contain any missing values, so it is suggested to keep it as default. Default: True.
- **rel_thres** (float) - Optional - Relevance threshold, above which a sample is considered rare. Must be a real number between 0 and 1 (0, 1]. Default: 0.5.
- **rel_method** (str) - Optional - Method to define the relevance function, either `auto` or `manual`. If `manual`, must specify `rel_ctrl_pts_rg`. Default: 'auto'.
- **rel_xtrm_type** (str) - Optional - Distribution focus, `high`, `low`, or `both`. If `high`, rare cases having small y values will be considered as normal, and vice versa. Default: 'both'.
- **rel_coef** (float) - Optional - Coefficient for box plot. Default: 1.5.
- **rel_ctrl_pts_rg** (2D array) - Optional - Manually specify the regions of interest. See SMOGN advanced example for more details. Default: None.

### Returns
- **Pandas dataframe** - Re-sampled dataset.

### Raises
- **ValueError** - If an input attribute has wrong data type or invalid value, or relevance values are all zero or all one, or synthetic data contains missing values.

### Example
```python
>>> from ImbalancedLearningRegression import smote
>>> housing = pandas.read_csv("https://raw.githubusercontent.com/paobranco/ImbalancedLearningRegression/master/data/housing.csv")
>>> housing_smote = smote(data = housing, y = "SalePrice")
```
```

--------------------------------

### tomeklinks()

Source: https://imbalancedlearningregression.readthedocs.io/en/latest/tomeklinks.html

Documentation for the tomeklinks function, an under-sampling method that removes TomekLinks from the dataset.

```APIDOC
## tomeklinks(data, y, option='majority', drop_na_col=True, drop_na_row=True, rel_thres=0.5, rel_method='auto', rel_xtrm_type='both', rel_coef=1.5, rel_ctrl_pts_rg=None)

### Description
TomekLinks is an under-sampling method that under-samples the majority/minority/both class(es) by removing TomekLinks. This function returns a re-sampled dataset.

### Method
Function call

### Endpoint
`tomeklinks(data, y, option='majority', drop_na_col=True, drop_na_row=True, rel_thres=0.5, rel_method='auto', rel_xtrm_type='both', rel_coef=1.5, rel_ctrl_pts_rg=None)`

### Parameters
#### Path Parameters
- **data** (Pandas dataframe) - Required - Pandas dataframe, the dataset to re-sample.
- **y** (str) - Required - Column name of the target variable in the Pandas dataframe.
- **option** (str) - Optional - Sampling information to sample the data set. If `majority`, resample only the majority class; if `minority`, resample only the minority class; if `both`, resample both majority and minority class.
- **drop_na_col** (bool) - Optional - Determine whether or not automatically drop columns containing NaN values. The data frame should not contain any missing values, so it is suggested to keep it as default.
- **drop_na_row** (bool) - Optional - Determine whether or not automatically drop rows containing NaN values. The data frame should not contain any missing values, so it is suggested to keep it as default.
- **rel_thres** (float) - Optional - Relevance threshold, above which a sample is considered rare. Must be a real number between 0 and 1 (0, 1].
- **rel_method** (str) - Optional - Method to define the relevance function, either `auto` or `manual`. If `manual`, must specify `rel_ctrl_pts_rg`.
- **rel_xtrm_type** (str) - Optional - Distribution focus, `high`, `low`, or `both`. If `high`, rare cases having small y values will be considerd as normal, and vise versa.
- **rel_coef** (float) - Optional - Coefficient for box plot.
- **rel_ctrl_pts_rg** (2D array) - Optional - Manually specify the regions of interest. See SMOGN advanced example for more details.

### Request Example
```python
>>> from ImbalancedLearningRegression import tomeklinks
>>> housing = pandas.read_csv("https://raw.githubusercontent.com/paobranco/ImbalancedLearningRegression/master/data/housing.csv")
>>> housing_tomeklinks = tomeklinks(data = housing, y = "SalePrice")
```

### Response
#### Success Response
- **Returns** (Pandas dataframe) - Re-sampled dataset.

#### Response Example
```python
housing_tomeklinks
```

### Error Handling
- **ValueError** - If an input attribute has wrong data type or invalid value, or relevance values are all zero or all one, or synthetic data contains missing values.
```

--------------------------------

### gn()

Source: https://imbalancedlearningregression.readthedocs.io/en/latest/gn.html

Over-sampling method that synthesizes new samples by introducing Gaussian noise on the numeric attributes and target variables of the seed samples. It can optionally perform random under-sampling.

```APIDOC
## gn(data, y, pert=0.02, samp_method='balance', under_samp=True, drop_na_col=True, drop_na_row=True, replace=False, manual_perc=False, perc_u=-1, perc_o=-1, rel_thres=0.5, rel_method='auto', rel_xtrm_type='both', rel_coef=1.5, rel_ctrl_pts_rg=None)

### Description
Over-sampling method that synthesizes new samples by introducing small perturbations on the numeric attributes and target variables of the seed samples. This method has an optional choice of random under-sampling.

### Parameters
- **data** (Pandas dataframe) - Required - Pandas dataframe, the dataset to re-sample.
- **y** (str) - Required - Column name of the target variable in the Pandas dataframe.
- **pert** (float) - Optional - Perturbation amplitude. Must be a real number between 0 and 1 (0, 1]. Default: 0.02.
- **samp_method** (str) - Optional - Method to determine re-sampling percentage. Either 'balance' or 'extreme'. Default: 'balance'.
- **under_samp** (bool) - Optional - If True, random under-sampling will be conducted on the normal bins. Default: True.
- **drop_na_col** (bool) - Optional - Determine whether or not automatically drop columns containing NaN values. The data frame should not contain any missing values, so it is suggested to keep it as default. Default: True.
- **drop_na_row** (bool) - Optional - Determine whether or not automatically drop rows containing NaN values. The data frame should not contain any missing values, so it is suggested to keep it as default. Default: True.
- **replace** (bool) - Optional - For decimal part of the over-sampling percentage, a subset of original dataset will be chosen as base samples to introduce noise, the selection can be with or without replacement. Default: False.
- **manual_perc** (bool) - Optional - Keep the same percentage of re-sampling for all bins. If True, perc_u is required to be a real number between 0 and 1 (0, 1), and perc_o is required to be a positive real number. Default: False.
- **perc_u** (float) - Optional - User-specified fixed percentage of under-sampling for all bins. Must be a real number between 0 and 1 (0, 1) if manual_perc = True. Default: -1.
- **perc_o** (float) - Optional - User-specified fixed percentage of over-sampling for all bins. Must be a positive real number if manual_perc = True. Default: -1.
- **rel_thres** (float) - Optional - Relevance threshold, above which a sample is considered rare. Must be a real number between 0 and 1 (0, 1]. Default: 0.5.
- **rel_method** (str) - Optional - Method to define the relevance function, either 'auto' or 'manual'. If 'manual', must specify rel_ctrl_pts_rg. Default: 'auto'.
- **rel_xtrm_type** (str) - Optional - Distribution focus, 'high', 'low', or 'both'. If 'high', rare cases having small y values will be considered as normal, and vice versa. Default: 'both'.
- **rel_coef** (float) - Optional - Coefficient for box plot. Default: 1.5.
- **rel_ctrl_pts_rg** (2D array) - Optional - Manually specify the regions of interest. See SMOGN advanced example for more details. Default: None.

### Returns
- **Pandas dataframe** - Re-sampled dataset.

### Raises
- **ValueError** - If an input attribute has wrong data type or invalid value, or relevance values are all zero or all one, or synthetic data contains missing values.

### Example
```python
>>> from ImbalancedLearningRegression import gn
>>> housing = pandas.read_csv("https://raw.githubusercontent.com/paobranco/ImbalancedLearningRegression/master/data/housing.csv")
>>> housing_gn = gn(data = housing, y = "SalePrice"")
```

--------------------------------

### ro()

Source: https://imbalancedlearningregression.readthedocs.io/en/latest/ro.html

Random Over-sampling method that synthesizes new samples by randomly copying minority samples. It takes a pandas dataframe and target column name, and returns a re-sampled dataframe.

```APIDOC
## ro(data, y, samp_method='balance', drop_na_col=True, drop_na_row=True, replace=True, manual_perc=False, perc_o=-1, rel_thres=0.5, rel_method='auto', rel_xtrm_type='both', rel_coef=1.5, rel_ctrl_pts_rg=None)

### Description
Random Over-sampling is an over-sampling method that synthesizes new samples by randomly copying minority samples. This function re-samples the dataset to balance the target variable distribution.

### Parameters
- **data** (Pandas dataframe) - Required - The dataset to re-sample.
- **y** (str) - Required - Column name of the target variable in the Pandas dataframe.
- **samp_method** (str) - Optional - Method to determine re-sampling percentage. Either `balance` or `extreme`.
- **drop_na_col** (bool) - Optional - Determine whether or not automatically drop columns containing NaN values. The data frame should not contain any missing values, so it is suggested to keep it as default.
- **drop_na_row** (bool) - Optional - Determine whether or not automatically drop rows containing NaN values. The data frame should not contain any missing values, so it is suggested to keep it as default.
- **replace** (bool) - Optional - Randomly select sample to duplicate: with or without replacement.
- **manual_perc** (bool) - Optional - Keep the same percentage of re-sampling for all bins. If `True`, `perc_o` is required to be a positive real number.
- **perc_o** (float) - Optional - User-specified fixed percentage of over-sampling for all bins. Must be a positive real number if `manual_perc = True`.
- **rel_thres** (float) - Optional - Relevance threshold, above which a sample is considered rare. Must be a real number between 0 and 1 (0, 1].".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".".
- **rel_method** (str) - Optional - Method to define the relevance function, either `auto` or `manual`. If `manual`, must specify `rel_ctrl_pts_rg`.
- **rel_xtrm_type** (str) - Optional - Distribution focus, `high`, `low`, or `both`. If `high`, rare cases having small y values will be considerd as normal, and vise versa.
- **rel_coef** (float) - Optional - Coefficient for box plot.
- **rel_ctrl_pts_rg** (2D array) - Optional - Manually specify the regions of interest. See SMOGN advanced example for more details.

### Returns
- **Pandas dataframe** - Re-sampled dataset.

### Raises
- **ValueError** - If an input attribute has wrong data type or invalid value, or relevance values are all zero or all one, or synthetic data contains missing values.

### Example
```python
>>> from ImbalancedLearningRegression import ro
>>> housing = pandas.read_csv("https://raw.githubusercontent.com/paobranco/ImbalancedLearningRegression/master/data/housing.csv")
>>> housing_ro = ro(data = housing, y = "SalePrice")
```
```

--------------------------------

### enn(data, y, samp_method='balance', drop_na_col=True, drop_na_row=True, rel_thres=0.5, rel_method='auto', rel_xtrm_type='both', rel_coef=1.5, rel_ctrl_pts_rg=None, k=3, n_jobs=1, k_neighbors_classifier=None)

Source: https://imbalancedlearningregression.readthedocs.io/en/latest/enn.html

Performs Edited Nearest Neighbor under-sampling on a dataset. It removes majority samples that are misclassified by their k nearest neighbors, returning a re-sampled Pandas dataframe.

```APIDOC
## enn(data, y, samp_method='balance', drop_na_col=True, drop_na_row=True, rel_thres=0.5, rel_method='auto', rel_xtrm_type='both', rel_coef=1.5, rel_ctrl_pts_rg=None, k=3, n_jobs=1, k_neighbors_classifier=None)

### Description
Edited Nearest Neighbor is an under-sampling method that edits the majority set by removing some of the majority samples from the original majority set.

### Method
Function call

### Endpoint
enn()

### Parameters
- **data** (Pandas dataframe) - Required - Pandas dataframe, the dataset to re-sample.
- **y** (str) - Required - Column name of the target variable in the Pandas dataframe.
- **samp_method** (str) - Optional - Method to determine re-sampling percentage. Either `balance` or `extreme`.
- **drop_na_col** (bool) - Optional - Determine whether or not automatically drop columns containing NaN values. The data frame should not contain any missing values, so it is suggested to keep it as default.
- **drop_na_row** (bool) - Optional - Determine whether or not automatically drop rows containing NaN values. The data frame should not contain any missing values, so it is suggested to keep it as default.
- **rel_thres** (float) - Optional - Relevance threshold, above which a sample is considered rare. Must be a real number between 0 and 1 (0, 1].
- **rel_method** (str) - Optional - Method to define the relevance function, either `auto` or `manual`. If `manual`, must specify `rel_ctrl_pts_rg`.
- **rel_xtrm_type** (str) - Optional - Distribution focus, `high`, `low`, or `both`. If `high`, rare cases having small y values will be considerd as normal, and vise versa.
- **rel_coef** (float) - Optional - Coefficient for box plot.
- **rel_ctrl_pts_rg** (2D array) - Optional - Manually specify the regions of interest. See SMOGN advanced example for more details.
- **k** (int) - Optional - The number of neighbors considered. Must be a positive integer.
- **n_jobs** (int) - Optional - The number of parallel jobs to run for neighbors search. Must be an integer. See sklearn.neighbors.KNeighborsClassifier for more details.
- **k_neighbors_classifier** (KNeighborsClassifier) - Optional - If users want to define more parameters of KNeighborsClassifier, such as `weights`, `algorithm`, `leaf_size`, and `metric`, they can create an instance of KNeighborsClassifier and pass it to this method. In that case, setting `k` and `n_jobs` will have no effect.

### Returns
Re-sampled dataset.

### Return Type
Pandas dataframe

### Raises
- **ValueError** - If an input attribute has wrong data type or invalid value, or relevance values are all zero or all one.

### Example
```python
>>> from ImbalancedLearningRegression import enn
>>> housing = pandas.read_csv("https://raw.githubusercontent.com/paobranco/ImbalancedLearningRegression/master/data/housing.csv")
>>> housing_enn = enn(data = housing, y = "SalePrice")
```
```

--------------------------------

### adasyn(data, y, k=5, samp_method='balance', drop_na_col=True, drop_na_row=True, rel_thres=0.5, rel_method='auto', rel_xtrm_type='both', rel_coef=1.5, rel_ctrl_pts_rg=None)

Source: https://imbalancedlearningregression.readthedocs.io/en/latest/adasyn.html

The adasyn() function is an over-sampling method that generates synthetic samples for imbalanced regression datasets. It takes a Pandas dataframe and the target column name, and returns a re-sampled dataframe.

```APIDOC
## adasyn(data, y, k=5, samp_method='balance', drop_na_col=True, drop_na_row=True, rel_thres=0.5, rel_method='auto', rel_xtrm_type='both', rel_coef=1.5, rel_ctrl_pts_rg=None)

### Description
ADASYN is an over-sampling method that is similar to SMOTE, but more samples are generated for seed samples that are difficult to be learned.

### Parameters
- **data** (Pandas dataframe) - Required - Pandas dataframe, the dataset to re-sample.
- **y** (str) - Required - Column name of the target variable in the Pandas dataframe.
- **k** (int) - Optional - The number of neighbors considered. Must be a positive integer. Default: 5.
- **samp_method** (str) - Optional - Method to determine re-sampling percentage. Either `balance` or `extreme`. Default: 'balance'.
- **drop_na_col** (bool) - Optional - Determine whether or not automatically drop columns containing NaN values. The data frame should not contain any missing values, so it is suggested to keep it as default. Default: True.
- **drop_na_row** (bool) - Optional - Determine whether or not automatically drop rows containing NaN values. The data frame should not contain any missing values, so it is suggested to keep it as default. Default: True.
- **rel_thres** (float) - Optional - Relevance threshold, above which a sample is considered rare. Must be a real number between 0 and 1 (0, 1]. Default: 0.5.
- **rel_method** (str) - Optional - Method to define the relevance function, either `auto` or `manual`. If `manual`, must specify `rel_ctrl_pts_rg`. Default: 'auto'.
- **rel_xtrm_type** (str) - Optional - Distribution focus, `high`, `low`, or `both`. If `high`, rare cases having small y values will be considerd as normal, and vise versa. Default: 'both'.
- **rel_coef** (float) - Optional - Coefficient for box plot. Default: 1.5.
- **rel_ctrl_pts_rg** (2D array) - Optional - Manually specify the regions of interest. See SMOGN advanced example for more details. Default: None.

### Returns
- **Pandas dataframe** - Re-sampled dataset.

### Raises
- **ValueError** - If an input attribute has wrong data type or invalid value, or relevance values are all zero or all one, or synthetic data contains missing values.
- **AssertionError** - If normalized ratio ri does not sum up to one.

### Example
```python
>>> from ImbalancedLearningRegression import adasyn
>>> housing = pandas.read_csv("https://raw.githubusercontent.com/paobranco/ImbalancedLearningRegression/master/data/housing.csv")
>>> housing_adasyn = adasyn(data = housing, y = "SalePrice")
```
```
Tokens:
Raw