chart-tools 0.1.3

Powerful visualizations, and an easy to use, interactive api for exploring and loading datasets

chart-tools    

Install & Use

Must have python 3.9+

pip install chart-tools

import chart_tools as ct
ct.load_data() # outputs available pre-defined data sources

Data Interface

Easily load datasets and explore available sources with one line of code

• The load_data() function and DataSource object use Github's API to explore file structures in repositories containing .csv files, and easily load files into dataframes. Chart-tools has a pre-defined library (collection of repositories) for you to explore within your notebook and load data from.

Robust caching system designed for Python notebooks, performing great with large datasets.

• Any dataframe you load gets cached in memory, remembering which pandas keyword arguments you used when loading the file. Next time you load it, you'll get a copy of the cached dataframe, unless you pass different keyword arguments. Not only is this great for performance with large datasets, but it also eliminates the common need to declare a df_raw = ... and then use df = df_raw.copy() to get your original data again.

Has a pre-defined library of data sources to explore, and lets you easily define your own library

• Save an entire Github repository file structure (csv files only) to your desktop

Charts & Visualization

superheat

• A "super" correlation heatmap you can't find elsewhere, designed for speed and ease of use.
• Marks are sized dynamically based on correlation strength, drawing your eyes straight to the most important relationships.
• Easily filter out variables whose coefficients average below a threshold, or simply mask/hide marks below a threshold.

Examples