Origin of OkraHealth and Pre-Alpha Release
March 28, 2019
OkraHealth recognized a need for interpretable software health.
OkraHealth came out of a project-based course I took while finishing my MS in Data Science at Northeastern University. DS 6050 - Expeditions in Data Science was being piloted as an alternative to the core capstone class DS 5500, Information Visualization, in Spring 2019 only. The DS 6050 course encouraged students to solve real-world data science problems by applying the skills they obtained in previous classes of their Data Science program. Successful completion of the DS 6050 course required practical experience with key steps of any Data Science project. All projects were individual. An expert panel from industry reviewed each of the projects and provided feedback.
DS 6050 Course Structure and Requirements
The projects focused on analyzing a large dataset extracted from GitHub. This dataset required each student to overcome challenges related to large scale data acquisition (the target was 1 million projects), data cleaning, data representation (how the data is modeled), data storage (define a storage format and location using a relational database), how to analyze the data, and choosing machine learning techniques, as well as visualizing the results. The Data Science practices discussed in previous classes can be summarized as follows (adopted from Prof. Jan Vitek and “The levels of data science class” by Jeff Leek):
- Asking: How to define a question, identify relevant data, and design the experiment.
- Telling: Writing about data science, interpreting model figures and results.
- Practicing: Implementation skills required to perform the analysis.
- Scaling: Figuring out how to deal with large real world datasets.
- Solving: Use real data examples, often small, working through them as a case study.
- Science: Formulate your defined question, then solve it at scale.
All work from the project was submitted as an open source project in a GitHub repository. My classwork can be reviewed in github.com/tbonza/EDS19. A more in depth reading of my findings are available for review in my DS 6050 project report. You can also review the initial Okra Python package if it’s of interest.
US PyCon 2019 Sprints
PyCon is a the largest annual gathering for the community using and developing the open-source Python programming language. US PyCon 2019 consists of five events in a row: (1) tutorial days, (2) conference days, (3) development sprints, (4) summits, (5) and a job fair. Development sprints help advance your favorite open source project. Space and infrastructure is provided and all experience levels are welcome. OkraHealth was initially migrated from a DS 6050 class project to an open source project during the development sprints at US PyCon 2019.
Releasing a Pre-Alpha Version
A pre-Alpha version of OkraHealth is scheduled to be released by June 27th, 2019. This release is scheduled to be presented at the Django Boston Meetup Group. We anticipate the pre-Alpha version having the following features:
- Interpretable software health analytics for OPENedX and its corresponding Python dependencies.
- RESTful API for consumption of the software health analytics (Okra-API).
- Open source statistical library used for the software health analytics and data pipeline
(Okra-Core).
Please direct any feature enhancements or issues to the corresponding GitHub issues pages for each of the linked, Okra-API and Okra-Core, repositories.