SynapCores AutoML Guide

Build powerful machine learning models directly in SQL without writing any Python code.

Overview

SynapCores AutoML provides comprehensive options for creating machine learning experiments through SQL syntax. Train, tune, and deploy production-ready models using familiar database commands.

Task Types

Creating AutoML Experiments

Basic Syntax

Option 1: AS Syntax

CREATE EXPERIMENT <experiment_name> AS
<SELECT_query>
WITH (<options>)

Option 2: USING Syntax

CREATE EXPERIMENT <experiment_name>
USING (<SELECT_query>)
TARGET <target_column>
OPTIONS (<options>)

Configuration Options

General Options

Available Algorithms

• 'linear_regression' - Linear Regression
• 'logistic_regression' - Logistic Regression
• 'decision_tree' - Decision Tree
• 'random_forest' - Random Forest
• 'gradient_boosting' - Gradient Boosting
• 'xgboost' - XGBoost
• 'neural_network' - Neural Network
• 'knn' - K-Nearest Neighbors
• 'naive_bayes' - Naive Bayes
• 'svm' - Support Vector Machine

Algorithm Selection Strategies

• 'all' - Try all available algorithms
• 'fast' - Only fast algorithms (linear models, decision trees, naive bayes, knn)
• 'accurate' - Only highly accurate algorithms (random forest, gradient boosting, xgboost, neural networks)
• 'interpretable' - Only interpretable algorithms (linear regression, logistic regression, decision trees)

Algorithm-Specific Options

Random Forest

WITH (
  task_type='classification',
  algorithms=['random_forest'],
  n_estimators=200,
  max_depth=10
)

Neural Network

WITH (
  task_type='classification',
  algorithms=['neural_network'],
  hidden_layers=[128, 64, 32],
  dropout_rate=0.2
)

Gradient Boosting / XGBoost

Feature Engineering Options

Scaling Methods

• 'standard' - Standardization (zero mean, unit variance)
• 'minmax' - Min-Max scaling to [0, 1]
• 'robust' - Robust scaling using median and IQR
• 'none' - No scaling

Categorical Encoding

• 'onehot' - One-hot encoding
• 'label' - Label encoding
• 'target' - Target encoding
• 'ordinal' - Ordinal encoding

Complete Examples

Customer Churn Prediction

CREATE EXPERIMENT churn_prediction AS
SELECT customer_id, age, tenure, monthly_charges, total_charges, churned
FROM customers
WITH (
  task_type='binary_classification',
  target_column='churned',
  max_trials=50,
  validation_split=0.2
);

House Price Regression

CREATE EXPERIMENT house_price_model AS
SELECT * FROM housing_data
WITH (
  task_type='regression',
  target_column='price',
  algorithms=['random_forest', 'xgboost', 'gradient_boosting'],
  max_trials=100,
  n_estimators=200
);

Fraud Detection with Feature Engineering

CREATE EXPERIMENT fraud_detection AS
SELECT * FROM transactions
WITH (
  task_type='binary_classification',
  target_column='is_fraud',
  algorithms=['xgboost', 'neural_network'],
  auto_features=true,
  polynomial_degree=2,
  interaction_features=true,
  scaling='robust',
  categorical_encoding='target',
  max_trials=150
);

Time Series Forecasting

CREATE EXPERIMENT sales_forecast AS
SELECT date, product_id, sales, promotions, holidays
FROM sales_data
WITH (
  task_type='time_series',
  target_column='sales',
  algorithms=['gradient_boosting', 'neural_network'],
  cv_folds=5
);

Interpretable Model for Compliance

CREATE EXPERIMENT loan_approval AS
SELECT * FROM loan_applications
WITH (
  task_type='binary_classification',
  target_column='approved',
  algorithms=['logistic_regression', 'decision_tree'],
  max_depth=5
);

Model Operations

Show All Experiments

SHOW MODELS;

Deploy a Model

DEPLOY MODEL best_model FROM EXPERIMENT churn_prediction
WITH (replicas=3, memory='2Gi');

Make Predictions

PREDICT churn_probability, risk_score
USING churn_model
AS SELECT customer_id, age, tenure FROM new_customers;

Describe a Model

DESCRIBE MODEL churn_model;

Drop a Model

DROP MODEL old_model;

Best Practices

1. Parameter Tuning: Algorithm-specific options apply to all selected algorithms where compatible.

2. Default Values: All options have sensible defaults. Only specify options that differ from defaults.

3. Resource Limits: Experiments respect both max_trials and time_budget_minutes. Stops when either limit is reached.

4. Reproducibility: Set random_seed for consistent results across runs.

5. Algorithm Compatibility: The system automatically filters incompatible algorithms for each task type.

Document Version: 1.0
Last Updated: December 2025
Website: https://synapcores.com

Originally published at synapcores.com — SynapCores is a free, single-binary AI-native database (vector + graph + SQL + LLM).