Classification Modeling

What it does

This skill enables AI agents to build and evaluate classification models from data, predicting categories or labels with specified accuracy goals. It handles various classification types including binary (two classes), multiclass (more than two classes), and multi-label (multiple classes per observation). The skill utilizes common algorithms like Logistic Regression, Decision Trees, Random Forest, Gradient Boosting, SVM, and Naive Bayes to achieve these predictions.

When to use it

• Predicting binary outcomes such as customer churn or loan default.
• Classifying items into multiple categories, for example, product types or sentiment analysis.
• Building credit scoring models or risk assessment systems.
• Identifying disease diagnosis from patient data.
• Predicting customer purchase likelihood or response to marketing campaigns.
• Detecting fraud, anomalies, or quality defects in production systems.

Key capabilities

• Classification Types: Supports binary, multiclass, and multi-label classification.
• Algorithm Selection: Implements Logistic Regression, Decision Trees, Random Forest, Gradient Boosting, SVM, and Naive Bayes.
• Evaluation Metrics: Calculates accuracy, precision, recall, F1-score, and AUC-ROC to assess model performance.
• Python Implementation: Provides code examples using Python libraries like pandas, numpy, scikit-learn, and matplotlib for data manipulation, modeling, and visualization.

Example prompts

• "Build a classification model to predict customer churn based on this dataset."
• "Classify these product descriptions into categories: [list of descriptions]."
• "Evaluate the performance of a logistic regression model using precision and recall metrics."

Tips & gotchas

• The skill uses Python and associated libraries (scikit-learn, pandas, numpy). Ensure the agent has access to a Python environment with these dependencies installed.
• Data standardization is performed as part of the implementation; this may impact results depending on the nature of your data.