Data science consists of extremely time-consuming and complex tasks. Luckily, there are a lot of open sources in Python that provide convenience for developers. Here, I would like to suggest some lesser-known libraries — they are not essential, but good choices for data science project perfection.
Cookiecutter: Better Project Template
Cookiecutter provides a better data science template for a user who wants to build a project. It can quickly organise a messy source code, files and data for the user.
pip install --user CookiecutterMore info: Cookiecutter, Github
Pandas-profilling: Speed Up Exploratory Data Analysis(EDA)
It can speed up the process of Exploratory Data Analysis (EDA) with only single line code. This includes missing values detection, distribution, correlation, etc.
## Using pip
pip install pandas-profiling
## Using Conda
conda install -c conda-forge pandas-profilingMore info: pandas-profilling, Github
imbalanced-learn: Handle imbalanced data
A professional library that can handle data with imbalanced labels, which includes techniques of resampling like: Over-sampling (SMOTE), Under-sampling (TomekLinks) and combine-sampling (SMOTEENN). It also provides an ensemble balanced model for machine learning.
## Using pip
pip install imbalanced-learn
## Using Conda
conda install -c conda-forge imbalanced-learnMore info: imbalanced-learn, Github
featuretools: Automated features engineering
Open source and tools automated features engineering.
## Using pip
pip install featuretools
## Using Conda
conda install -c conda-forge featuretoolsMore info: featuretools, Github
Cufflinks
An excellent data visualisation tool for easy interactive Pandas charting with Plotly for the users.
## Using pip
pip install cufflinks
## Using Conda
conda install -c conda-forge cufflinks-pyMore info: Github
Conclusions
Indeed, data science projects are time-consuming and complex, therefore these libraries are absolutely recommended for speeding up the process and reducing the bothersome procedures of developing a data science project. If you have any other open sources to share, you are welcome to list them out in your comment.
