https://github.com/akoury/ml-helper

Python library with helpers to speed up and structure machine learning projects.
https://github.com/akoury/ml-helper

data data-visualization machine-learning ml python scikit-learn sklearn

Last synced: 27 days ago
JSON representation

Python library with helpers to speed up and structure machine learning projects.

• Host: GitHub
• URL: https://github.com/akoury/ml-helper
• Owner: akoury
• License: mit
• Created: 2019-02-20T16:48:09.000Z (over 5 years ago)
• Default Branch: master
• Last Pushed: 2019-06-05T12:30:18.000Z (over 5 years ago)
• Last Synced: 2024-09-27T20:23:08.928Z (about 1 month ago)
• Topics: data, data-visualization, machine-learning, ml, python, scikit-learn, sklearn
• Language: Python
• Homepage: https://pypi.org/project/ml-helper/
• Size: 11.7 MB
• Stars: 10
• Watchers: 3
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: license.txt

Awesome Lists containing this project

README

        
# ML Helper

---

Helpers to speed up and structure machine learning projects.

The library is available in [Pypi](https://pypi.org/project/ml-helper/)

### Installing

---

The easiest way to install ml-helper is through ```pip```

```python

pip install ml-helper

```

To use it in your project, you must first import the library

```python

from ml_helper.helper import Helper

```

And then create a Helper object with a dictionary of keys related to your project

```python

KEYS = {

    'SEED': 1,

    'TARGET': 'y',

    'METRIC': 'r2',

    'TIMESERIES': True,

    'SPLITS': 5

}

hp = Helper(KEYS)

```

After this, you may use the helper object's many functions

#### Dependencies

ML-Helper requires:

* Python (>3.5)

* Numpy (>=1.16)

* Pandas (>=0.23.4)

* Seaborn (>=0.9)

* Scikit-learn (>=0.20)

* Natplotlib (>=3)

* Scipy (>=1)

* Imblearn

* Vecstack

### Functionality

---

The functionality is separated into 4 groups:

* Data Exploration

    * Missing Data

    * Boxplot of numerical variables

    * Coefficient of variation

    * Correlation (numerical and categorical)

    * Under Represented Features

    * Target Variable Distribution

    * Feature Importance

    * PCA Component Variance

* Data Preparation

    * Convert features to categories

    * Drop multiple columns

* Modeling

    * Cross Validation (with stratified kfolds, or time series split depending on use case)

        * Randomized Grid Search

    * Pipeline: Collection of models and pipeline steps that get performed and scored

    * Predict: Predict on unseen data

    * Stack Predict: Build a stacked model and perform a prediction

* Regression

    * Plots for predictions

* Classification

    * ROC Curve

    * Classification Report

* Others

    * Select features based on types

    * Split X and y

    * Plot models/pipelines

### Working Examples

---

If you wish to see the library in use, you may view the notebooks in the [examples](examples) section.

Also, you can see the implementation in their corresponding Kaggle Kernels:

* [Bike Sharing in Washington D.C.: Time Series Regression](https://www.kaggle.com/akoury/bike-sharing-in-washington-d-c-using-ml-helper)

* [Employee Attrition: Classification](https://www.kaggle.com/akoury/employee-attrition-basis-to-create-ml-helper-lib)

### ML-Helper Coding Style

---

Ml-Helper complies to PEP8 and uses ```black``` for coding standards

### Versioning

---

[SemVer](http://semver.org/) is used for versioning. 

### License

---

This project is licensed under the MIT License - see the [License](license.txt) file for details