Collection Structure

General Structure

On start-up, the MindsDB project consists of 2 collections: models and models_versions.

You can verify it by running the following MQL commands:

1. First, switch to the mindsdb project.

   > use mindsdb
   < 'switched to db mindsdb'
   
   Here is the doc page on the PROJECT entity to learn more.

2. Then, query for the available collections.

   > show collections
   < models
     models_versions
   

The models Collection

The models collection stores information about the models, such as name, status, accuracy, and more.

> db.models.find({})
< {
    NAME: 'home_rentals_model',
    ENGINE: 'lightwood',
    PROJECT: 'mindsdb',
    VERSION: 1,
    STATUS: 'complete',
    ACCURACY: 1,
    PREDICT: 'rental_price',
    UPDATE_STATUS: 'up_to_date',
    MINDSDB_VERSION: '23.2.4.0',
    ERROR: null,
    SELECT_DATA_QUERY: 'db.home_rentals.find({})',
    TRAINING_OPTIONS: "{'target': 'rental_price', 'using': {}}",
    TAG: null,
    CREATED_AT: 2023-02-24T16:05:43.248Z,
    _id: ObjectId("000000000000007725907968")
  }

Where:

Name	Description
NAME	The name of the model.
ENGINE	The engine used to create the model.
PROJECT	The project where the model resides.
VERSION	The version of the model.
STATUS	Training status (generating, or training, or complete, or error).
ACCURACY	The model accuracy (0.999 is a sample accuracy value).
PREDICT	The name of the target column to be predicted.
UPDATE_STATUS	Training update status (up_to_date, or updating, or available).
MINDSDB_VERSION	The MindsDB version used while training (22.8.2.1 is a sample version value).
ERROR	Error message stores a value in case of an error, otherwise, it is null.
SELECT_DATA_QUERY	It is a query that selects input data.
TRAINING_OPTIONS	Additional training parameters.
TAG	The models can have tags to classify them into categories.

Check out this doc page to learn how to create, view, and drop models.

The models_versions Collection

The models_versions collection stores information about all present and past versions of each model.

To learn more about managing models’ versions, follow our docs here.

> db.models_versions.find({})
< {
    NAME: 'home_rentals_model',
    ENGINE: 'lightwood',
    PROJECT: 'mindsdb',
    ACTIVE: true,
    VERSION: 1,
    STATUS: 'complete',
    ACCURACY: 1,
    PREDICT: 'rental_price',
    UPDATE_STATUS: 'up_to_date',
    MINDSDB_VERSION: '23.2.4.0',
    ERROR: null,
    SELECT_DATA_QUERY: 'db.home_rentals.find({})',
    TRAINING_OPTIONS: "{'target': 'rental_price', 'using': {}}",
    TAG: null,
    CREATED_AT: 2023-02-24T16:05:43.248Z,
    _id: ObjectId("000000000000007725907969")
  }

Where:

Name	Description
NAME	The name of the model.
ENGINE	The engine used to create the model.
PROJECT	The project where the model resides.
ACTIVE	The status of the model’s version. The value of true if active, the value fo false if inactive.
VERSION	The version of the model.
STATUS	Training status (generating, or training, or complete, or error).
ACCURACY	The model accuracy (0.999 is a sample accuracy value).
PREDICT	The name of the target column to be predicted.
UPDATE_STATUS	Training update status (up_to_date, or updating, or available).
MINDSDB_VERSION	The MindsDB version used while training (22.8.2.1 is a sample version value).
ERROR	Error message stores a value in case of an error, otherwise, it is null.
SELECT_DATA_QUERY	It is a query that selects input data.
TRAINING_OPTIONS	Additional training parameters.
TAG	The models can have tags to classify them into categories.

Check out this doc page to learn how to work with models’ versions.

To view all databases/projects, use the command below.

> show databases
< admin               64.00 KiB
  information_schema  64.00 KiB
  mindsdb             64.00 KiB
  files               64.00 KiB
  mongo_demo_db       64.00 KiB

Please note that mindsdb is the default project. Here is the doc page on the PROJECT entity to learn more.

Please follow the MindsDB Information Schema doc page to learn more.