Skip to content

Query the model from MariaDB database

Prerequisite

Don't forget to enable CONNECT Storage Engine as explained in connect your data section.

This section assumes that you have trained a new model using MariaDB or MindsDB Studio. To query the model, you will need to SELECT from the model table:

SELECT
  <target_variable> AS predicted,
  <target_variable_confidence> AS confidence,
  <target_variable_explain> AS info
FROM
  mindsdb.<model_name>
WHERE
  <feature_one> AND <feature_two>

Query the model from other databases

Note that even if you have trained the model from the MariaDB database, you will be able to query it from other databases too.

Query example

The following example shows you how to train a new model from the MariaDB client. The table used for training the model is the Used cars dataset. MindsDB will predict the price based on the values added in the WHERE clause.

SELECT
  price AS predicted,
  price_confidence AS confidence,
  price_explain AS info
FROM
  mindsdb.used_cars_model
WHERE
  model = "A6"
  AND mileage = 36203
  AND transmission = "Automatic"
  AND fuelType = "Diesel"
  AND mpg = "64.2"
  AND engineSize = 2
  AND year = 2016
  AND tax = 20;

You should get a response from MindsDB similar to:

price confidence info
16117 0.98 Check JSON below 
info: {
   "predicted_value": 16117.627834024992,
   "confidence": 0.98,
   "prediction_quality": "very confident",
   "confidence_interval": [10737.135673357996, 21498.119994691988],
   "important_missing_information": []
}

Model predictions

Comments