FAQ
What is the roadmap?
MindsDB roadmap is aimed to be aligned with our goals:
Success
- versions 1.0
- MindsDB GUI where you can visualize explainability goals
- Support for images and complex text
- MindsDB REST API's
Success
- versions 2.0
- Run Machine Learning Models as Tables
Important
- versions 3.0
- Auto ETL and Data Preparation
What type of data can MindsDB learn and predict from?
We support tabular data fromats as a CSV, Excel, JSON, text files also pandas data frame, URLs, s3 files. From the database integrations we support:
| SQL | NoSQL | Streams | Data Warehouse |
|---|---|---|---|
 |
 |
 |
 |
 |
 |
 |
- |
 |
- | - | - |
 |
- | - | - |
 |
- | - | - |
 |
- | - | - |
 |
- | - | - |
 |
- | - | - |
How does it work?
In very simple terms, MindsDB follows the following steps:
- to learn:
- break data source intro train, test, validate
- transform data source into tensors
- build and train encoders (if necessary)
- produce a neural network based model that can take in the input tensor and produce the target tensor
- break train data into batches and try learning a model that can fit the target
- test and validate until model convergence
- store metadata about the most fit model
- to predict:
- transform question data into input tensor
- load most fit model
- run input tensor into model
- transform output tensor into readable output
How can I help?
You can help in the following ways:
- Trying MindsDB and reporting issues.
- If you know python, you can also help us debug open issues. Issues labels with the
good first issuetag should be the easiest to start with. - You can help us with documentation and examples.
- Tell your friends, write a blog post about MindsDB.
- Join our team, we are growing fast so we should have a few open positions.
Why is it called MindsDB?
Well, as most names, we needed one, we like science fiction and the culture series, where there are these AI super smart entities called Minds.
How about the DB part?. Although in the future we will support all kinds of data, currently our objective is to add intelligence to existing data stores/databases, hence the term DB. As to becoming a Mind to your DB.
Why the bear? Who doesn't like bears! Anyway, a bear for UC Berkeley where this all was initially coded.