Summary
Erdre is a machine learning pipeline that creates models for time series data. Taking any type of time series data as input, the pipeline can be used to produce machine learning models for regression, classification, and future prediction. The tool also provides ways of visualizing and evaluating the results, in addition to an API for querying inference on the models. Erdre can be configured to use any of the most common machine learning algorithms, such as neural networks (fully connected, convolutional, recurrent), decision trees, random forests and boosting methods.
The features of the machine learning pipeline consist of the following:
• High configurability
• Data profiling and cleaning
• Feature engineering
• Made for both regression and classification
• Choose and adjust machine learning algorithm
• Web UI for creating models
• Command Line Interface for creating models
• API for interacting with models, which can be run both locally and deployed to the cloud.
• Powered by Data Version Control (DVC), which gives robust control over the configuration parameters and input data used to create each model.
More information & hyperlinks
| Web resources: | https://www.sintef.no/en/expertise/digital/sustainable-communication-technologies/trustworthy-green-iot-software/ |
| Country: | NO |
| Address: | SINTEF, Forskningsveien 1, Oslo 373 |
Video
Search
Contacts
Not specified (see website if available) or see associated project
Organisations
