This project is based on MindsDB (https://github.com/mindsdb/mindsdb) and MindsDB_sql (https://github.com/mindsdb/mindsdb_sql). The Github repo of our project is at https://github.com/talhz/mindsdb_recommend.

Instructions related to our new syntax

conda create -n py39 python=3.9
conda activate py39

cd mindsdb
pip3 install -e .
pip install mindsdb[lightwood]
cd ..

cd mindsdb_sql
pip3 install -e .
cd ..

cd mindsdb
python -m mindsdb

To exit from virtual environment and stop all

conda deactivate

The following query will recommend the suitable model types of the selecting data.

RECOMMEND MODEL project_name.predictor_name
FROM integration_name
    (SELECT column_name, ... FROM table_name)
PREDICT target_column

• project_name.predictor_name is the name of model we want to create
• FROM ... is the syntax to get x (sample) data to be used in model
• PREDICT ... is the syntax to define y (label) data to be used in model

To check the recommendation result, simply use the following query:

DRSCRIBE project_name.predictor_name;

The recommendation result is given in the last column. By default, it provides the top 3 models with highest accuracy. Users can then choose the model at their own discretion.

from mindsdb_sql import parse_sql
from mindsdb.api.executor.command_executor import ExecuteCommands

query = parse_sql(sql, dialect="mindsdb")
ret = command_executor.execute_command(query)

• parse_sql: parse sql query to the mindsdb
• execute_command: execute the parsed results and return the output

Make sure that you are in the root directory.

python test.py

• mindsdb_sql\mindsdb_sql\parser\dialects\mindsdb\parser.py
• mindsdb_sql\mindsdb_sql\parser\dialects\mindsdb\lexer.py
• mindsdb_sql\mindsdb_sql\parser\ast\recommend.py
• mindsdb_sql\mindsdb_sql\parser\dialects\mindsdb\recommend_model.py

recommend.py and recommend_model.py are the places to interpret semantics and to output interpreted strings. To parse sql query to the mindsdb readable contents, edit recommend.py and recommend_model.py.

This is the code for parsing sql query to mindsdb

• mindsdb\mindsdb\api\mysql\mysql_proxy\executor\mysql_executor.py
• mindsdb\mindsdb\api\executor\command_executor.py

This is the code for executing the parsed results and return the execution output