The response to our launch of Loominus Public has been incredible! We are busy keeping up with all the invites and registering users as fast as possible. While our public beta program continues, we love that our users help shape the future of Teraport and Learner. This post describes some of the areas where our users contribute.
Loominus has three pricing plans (more details to be published soon):
- Public – Free to with shared data and models
- Teams – Data and models private to your team
- Enterprise – Dedicated infrastructure and support
We listen to our users and encourage everyone to use the platform and interact with us as you find bugs or have feature ideas.
To get a rundown how the Loominus platform is used, watch an eight minute video to learn how to use the platform end-to-end.
Help Shape Teraport
Teraport allows users to feature engineer and transform data in two main pipelines: the API and reporting tables.
When an input data stream is created, the data transforms as it passes through the API. These transformations are Preprocessors and Filters.
Preprocessors allow file types such as CSV, XLSX and PDF be adapted to the low level JSON REST API.
The preprocessor framework is extensible so drop us a line if you have additional preprocessors you need.
Filters transform data as it passes through the Teraport data ingestion JSON REST API. Filters are used for feature engineering.
Examples of this include:
- ISO Date Normalizer – Detects temporal fields in your data stream and normalize values to ISO date format
- Drop Rows – Drops data with missing values
- Privacy – Anonymizes sensitive fields in your data stream (Teraport will not store the original values of the anonymized field)
The filter and transform framework is extensible so drop us a line if you have additional ones you need.
As Teraport ingests data, it’s staged into Teraport’s internal staging tables. These staging tables are sourced to define reporting tables using the Teraport UI. The Table Designer is where users apply various transformations to build new features.
Examples of these include:
- SQL Select – Uses SQL Syntax to filter and transform
- Pivot – Summarizes the data by reshaping it from long to wide format.
- Subset Columns – Projects selected columns
Teraport has about 20 transforms. Check out the post on Building Data Pipelines with Teraport for all the details on how to apply various transformations.
The transformation framework is extensible so drop us a line if you have additional transformation you need.
Learner helps you get insights from your data without writing any code. Use the same analysis techniques and modeling frameworks the experts use to get world class results faster.
Canned analysis is a push-button approach to common data operations. It can be run over all the data within a Learner project or focused on a single target column.
Some of the general canned analysis include:
- Correlation analysis – Discover and explore correlations between the various columns in your data
- Outlier Analysis – Determine which columns have outliers, how many outliers and how removing changes the column value distributions
Examples of target canned analysis include:
- Feature interaction – Automatically determine which columns in a pair give the most predictive lift towards predicting a target column
- Rules analysis – Discover segments in your data to explain where the predictiveness is stronger or weaker towards a target column
- Decision tree analysis – Understand how data flows when your data and target column can be presented via an interactive decision tree
The canned analysis framework is extensible so drop us a line if you have additional push-button analyses you need.
Machine Learning Models
Learner was built using our Provider Plugin architecture. This means different machine learning frameworks can be integrated providing a common interface to help build state of the art models with no code.
We support the following frameworks:
The Provider Plugin architecture is extensible so drop us a line if you have additional frameworks you need.
In future posts we’ll dive deeper into the Learner workflow. You’ll learn how it can be used to experiment and build models iteratively to solve business problems.
Help shape the future of Teraport and Learner!
Use Loominus for Free
Help your business achieve machine learning success