Welcome to Hygia’s documentation#

Hygia is an open source, MIT-licensed library providing high-performance, easy-to-use playground to organize, compare, register and share Machine Learning experiments on Databases developed in Python programming language.

Why Hygia?#

1. Rework#

Data Scientists often invest significant time and effort in creating experiments, developing functions, and retraining models to identify Keyboard Smash. However, this process can be hindered by replication of functions and code. Reusing code with errors can lead to unreliable results, rendering all of the processing and study done using that code invalid. Therefore, it is essential to ensure that the code being reused is thoroughly tested and validated before it is used in any further experiments or analysis.

Alternative text

2. Loosing time and spend money#

The data science process can become repetitive and time-consuming when working with different data from various structures. The testing and experimentation required for each new dataset can add to the repetitive nature of the work. Additionally, maintaining and evolving the generated codes can pose significant challenges, especially when the necessary processes, such as storage, versioning, and evolution, are not given the necessary attention. This can have a detrimental long-term impact on the overall data science process.

Alternative text

3. Hygia as a solution#

Hygia offers a flexible solution for configuring your data pipeline. With it, you can customize the operations executed at each stage of the pipeline. Our current implementation includes the identification of Keyboard Smash for the Latin alphabet (also known as the Roman alphabet) through specific functions that have been integrated into the pipeline. This allows for efficient and accurate identification of Keyboard Smash in your data.

Alternative text

The main features are:

  • Be able to experiment on several databases

  • Customize data pipeline for each database

  • Custom visualization

  • Accelerate new hythoses evaluation

  • Search, debug, and compare experiments, datasets, and models

  • Organize and display experiments and model metadata however you want

  • Share and collaborate on experiment results and models across the org

4. Supported by#

Alternative text Alternative text Alternative text