API Reference

Complete reference for Scald classes and methods.

Scald

Main orchestrator for AutoML workflows.

Main orchestrator for Actor-Critic ML automation.

run async

run(
    train_path: str | Path,
    test_path: str | Path,
    target: str,
    task_type: TaskType,
) -> np.ndarray

Execute Actor-Critic loop with long-term memory.

Usage

import asyncio
from scald import Scald

async def main():
    scald = Scald(max_iterations=5)

    predictions = await scald.run(
        train_path="train.csv",
        test_path="test.csv",
        target="price",
        task_type="regression"
    )

    return predictions

results = asyncio.run(main())

Type Signatures

from typing import List

class Scald:
    def __init__(self, max_iterations: int = 5) -> None: ...

    async def run(
        self,
        train_path: str,
        test_path: str,
        target: str,
        task_type: str
    ) -> List[float | int | str]: ...

Return Values

run() returns predictions as a list. For classification, list contains class labels (int or str). For regression, list contains numeric predictions (float). List length matches test data row count.

Exceptions

Common exceptions: FileNotFoundError for missing data files, ValueError for invalid task_type or missing target column, RuntimeError for API or execution failures.

Error handling:

try:
    predictions = await scald.run(...)
except FileNotFoundError:
    print("Data file missing")
except ValueError:
    print("Invalid parameters")
except Exception as e:
    print(f"Error: {e}")

See Python API Guide for practical examples and Configuration for settings.