Most data scientists spend a significant amount of their time on prepwork to ensure their data is ready for modeling, and while data teams often have very similar goals, models are usually built from scratch because there is no common way to structure data.
Objectiv proposes the adoption of a common taxonomy for analytics so models & data can be reused and data scientists can build on knowledge and practises of others.
Objectiv is built around the open taxonomy of analytics, which is our proposal for a common way to collect, structure and validate data.
It defines classes for each common event type and the contexts in which they can happen. It describes their properties, requirements and their relationships with other classes.
With Objectiv, you map your application to this taxonomy, creating a contextual layer that's used for data collection & validation.
In most cases you can skip the discussion on what to track, because the taxonomy is designed to ensure the collected data covers a wide range of common analytics use cases.
The initial version of the taxonomy is built for product analytics. We have plans to support other fields as well.
Objectiv comes with a set of tools that help you set up error-free tracking instrumentation.
For instance, you can validate your instrumentation against the taxonomy and get live feedback in your IDE and console while developing, enabling you to catch errors before data starts flowing in.
Objectiv’s tracker collects data that is unusually rich. Events can carry multiple contexts, including the exact location in the UI from where they were triggered.
All data is well-structured, self-descriptive and has been validated at the first step of the pipeline, so you can use it for modeling with minimal additional gruntwork.
Objectiv features an SQL abstraction layer for modeling that enables you to use familiar dataframe operations on your full data set, straight from your notebook.
You can output your models to SQL queries with a single command, effectively closing the gap between experimentation and production.
By embracing the open taxonomy, you can reuse your own models for other projects and quickly build models reusing parts of others.
That retention model you’ve built for your campaign site? You can likely reuse that for your main site by changing a single line of code.