Machine Learning

Overview

The Machine Learning package provides a set of tools for implementing standard machine learning algorithms directly within your workflows. It enables you to train models and generate predictions using data currently held in the workflow's memory.

By integrating these bricks, you can add predictive logic—such as categorization, value forecasting, or pattern recognition—to your automation processes.

Important: In-Memory Processing & Data Limits

These bricks are designed for in-memory execution. This means all data processing, model training, and prediction occur within the active memory (RAM) allocated to the workflow.

• Suitable For: Small to medium-sized datasets.
• Not Suitable For: "Big Data" or massive datasets (millions of rows).

Advisory: Attempting to process datasets that exceed the available system memory will result in significant performance degradation or workflow failure. For massive datasets, it is recommended to perform heavy processing in a dedicated database or external data warehouse before passing summarized results to these bricks.

Capabilities

This package focuses on three primary areas of machine learning:

1. Classification

Used to predict a category or label for a given input. The model learns from labeled historical data to assign classes to new items.

• Use Cases: Spam detection, categorizing support tickets, or binary decisions (e.g., approve/reject).

2. Regression

Used to predict continuous numerical values. The model analyzes the relationship between input variables and a target number.

• Use Cases: Forecasting sales figures, estimating costs, or predicting durations.

3. Clustering

Used to group data points based on inherent similarities without relying on predefined labels.

• Use Cases: Customer segmentation, anomaly detection, or organizing unstructured data.