dataramp 1.0.1.dev181

A Data science library for data science / data analysis teams

Dataramp

Dataramp is a Python library designed to streamline data science and data analysis workflows. It offers a collection of utility functions and tools tailored to assist data science teams in various aspects of their projects.

Key Features

1. Project Management

• Simplify project setup with a single function call to generate a standardized project directory structure.
• Organize datasets, model outputs, scripts, notebooks, and more in predefined folders for better project management.

2. Model Saving and Loading

• Save and load trained machine learning models effortlessly.
• Supports multiple formats including joblib, pickle, and keras for compatibility with diverse model types.

3. Data Exploration and Visualization

• Explore datasets and generate summary statistics with ease.
• Visualize feature distributions and missing data patterns to gain insights into your data.

4. Feature Engineering

• Handle missing data and outliers effectively.
• Drop missing columns based on user-defined thresholds and detect outliers using Tukey's Interquartile Range (IQR) method.

5. Model Evaluation and Cross-Validation

• Evaluate model performance with comprehensive metrics such as accuracy, F1-score, precision, and recall.
• Generate classification reports and support cross-validation for robust model evaluation.

6. Scaling and Normalization

• Scale and normalize data using min-max scaling and z-score normalization techniques.
• Bring features to a common scale for improved model performance.

By providing a range of functionalities, Dataramp aims to enhance productivity and efficiency in data science projects, empowering teams to focus on deriving meaningful insights from their data.

Quickstart

To get started with Dataramp in your data science projects, follow these simple steps:

You can install Dataramp via pip:

pip install dataramp 

To upgrade an existing installation of Dataramp, use:

pip install --upgrade dataramp

Getting Started

Once installed, you can import the library and explore its functionality:

import dataramp as dr

Creating a New Project

To create a new project using Dataramp, run:

dr.core.create_project("project-name")

This will create a project with a structured directory layout to kickstart your project.

Project Directory Structure

project-name/
├── datasets
│   └── dataset.csv
├── outputs
│   └── models
├── README.md
└── src
    ├── notebooks
    │   └── notebook.ipynb
    └── scripts
        ├── ingest
        └── tests

Sample Usage

import dataramp as dr  # import the dataramp library
import pandas as pd

from dataramp.utils import (
    describe_df,
    get_cat_vars,
    feature_summary,
    display_missing,
    get_unique_counts,
)

df = pd.read_csv("data/iris.csv")  # load iris dataset

df.head() #  Snapshot of your df

missing = display_missing(df)
print(missing)

Project Links

• GitHub Repository: dataramp
• PyPI Package: dataramp