Product
In-Database Machine Learning
MindsDB extends the SQL language so that Machine Learning models can be created, queried and maintained as if they were database tables
New Machine Learning SQL Syntax
The CREATE PREDICTOR statement is used to train a new model with either Automated Machine Learning or SQL Declarative ML.
CREATE PREDICTOR mindsdb.[predictor_name]
FROM [integration_name]
(SELECT [column_name, ...] FROM [table_name])
PREDICT [target_column]
--Optional parameters
USING [parameter_key] = ['parameter_value']
SELECT [target_variable], [target_variable]_explain
FROM mindsdb.[predictor_name]
WHERE [column1]=[value] AND [column2]=[value];
Get predictions and their explanations by querying your models as if they were database tables using the standard SELECT statement.
Use the JOIN statement to make bulk or time-series predictions by combining rows from the database table and the model on a related column.
SELECT t.[column_name], p.[column_name] ...
FROM [integration_name].[table] AS t
JOIN mindsdb.[predictor_name] AS p
CREATE PREDICTOR mindsdb.[predictor_name]
FROM [integration_name]
(SELECT * FROM [table_name])
PREDICT [target_column]
ORDER BY [sequential_column]
GROUP BY [partition_column]
WINDOW [int]
HORIZON [int];
To train a time-series model, MindsDB provides additional statements. It can train a multivariate model grouped by several columns at once. For example, make sales predictions for all products in all stores.
Bring your own ML models to the Database
MindsDB enables you to import models you have created in other frameworks such as Pytorch, Tensorflow, etc and query predictions from inside your Database or BI tool.
Bring Predictions into your BI Tools
MindsDB is a simpler way to use existing data to predict future events. It speaks the SQL Wire Protocol and thus works with the majority of business intelligence tools.
MindsDB is a member of the Intel® Partner Alliance, integrates with the Intel® Distribution of OpenVINO™ Toolkit,
and works with all Cloud Service Providers.
