Skip to content

Making Predictions Guide

This guide covers how to use your trained Arc models to make predictions on new data.

Overview

Once you've trained and evaluated a model, you can use it for inference (making predictions). Arc makes this simple: 1. Load new data - Get data you want predictions for 2. Run prediction - Apply your trained model 3. View results - Access predictions via SQL

Quick Start

Make predictions on new data:

Use the diabetes_predictor model to make predictions on new_patients table

Or with the /ml predict command:

/ml predict --model diabetes_predictor
            --data new_patients
            --output predictions

Using the /ml predict Command

Basic Prediction

/ml predict --model <model_name>
            --data <input_table>

Predictions are saved to <model_name>_predictions table by default.

Custom Output Table

/ml predict --model diabetes_predictor
            --data new_patients
            --output patient_risk_scores

Prediction Workflow

Prepare Input Data: Ensure new data has the same features as training data with identical preprocessing.

Run Prediction: /ml predict --model my_model --data new_data

View Predictions: /sql SELECT * FROM my_model_predictions LIMIT 10

Understanding Prediction Outputs

Classification: Returns predictions with probabilities and predicted classes.

Regression: Returns predicted values.

Example: Batch Scoring

/ml predict --model churn_predictor
            --data all_customers
            --output churn_scores

Query results with /sql SELECT * FROM churn_scores WHERE churn_probability > 0.7

Next Steps