We have released exciting new features and improvements to the MindsDB products in May. Read on for all of the details!
The latest version of MindsDB is 1.17.9 and includes the following new features:
- A new ClickHouse data source. Now you can easily ingest data stored in ClickHouse into MindsDB.
- Improvements to the confusion matrix and confidence evaluations.
The latest changes in the Lightwood 0.25.0 version include:
- Fixed issues around training data on small datasets that contain less than 40 rows.
- Improvements to the model switching from CPU to GPU and vice versa.
- Minor bug fixes in the Image encoder.
- Improved our test suites to evaluate the missing columns predictions.
We have migrated the MindsDB and Lightwood documentation websites into docs.mindsdb.com. The new documentation website includes everything from installation guidelines, MindsDB tutorials to MindsDB Server and SDKs examples. We would love to hear your ideas for improving our docs, examples, and tutorials in our documentation repository.
Our free online course Machine Learning Basics Using Mindsdb is still available and open for new students. To enroll in the course visit education.mindsdb.com.
MindsDB Python SDK
We have released a new software development kit for Python that provides all of the MindsDB’s native functionalities through MindsDB Server. For additional information check out the SDK documentation and Python SDK repository.
The new feature available in the latest version of MindsDB Server (1.9.4) provides filtering and pagination options to the DataSources list API. It is published and available for download on PyPi.
MindsDB Covid19 Census application
The COVID-19 system that we have developed in the last months aims to anonymously gather data from specific locations instead of city/countrywide ones. You can help by completing the census at covid.mindsdb.com.
In the past few weeks, our team has implemented a lot of new features and visualization changes related to MindsDB Scout. The new data quality insights and model explainability features will make MindsDB Scout even more user-friendly. All of the mentioned features will be announced in the next few weeks, so stay tuned.
One of our main aims in the community is to make contributing to MindsDB as easy as it can be. Since MindsDB is designed to make it super easy for developers to use Machine Learning we would love to hear developers’ ideas, opinions and get contributions from the community. For further discussions join our community forum at github.com/mindsdb/mindsdb/discussions.