AutoML Analyst Skill

This skill turns the agent into a data scientist that autonomously trains models and reports results.

Workflow

1. •

   Data Analysis:

   • •Read the input file (CSV/JSON).
   • •Determine the problem type:
     • •Regression: Predicting a number (e.g., predicted_ltv).
     • •Classification: Predicting a category (e.g., churn/no_churn).
     • •Clustering: grouping data (e.g., customer_segments) if no target is specified.
2. •

   Code Generation:

   • •Write a python script (e.g., train_model.py) using pandas and scikit-learn.
   • •The script MUST output metrics and generate plots (saved as images).
3. •

   Reporting:

   • •Read resources/report_template.html.
   • •Inject the results (metrics, plot images, summary) into the template.
   • •Save the final report as report.html.

Code Guidelines

• •Handle missing values (impute or drop).
• •Encode categorical variables.
• •Use matplotlib or seaborn for plots.

Example

User: "Analyze this customer_churn.csv." Agent:

1. •Detects 'churn' column -> Classification problem.
2. •Writes train_churn.py -> Trains Random Forest -> Accuracy 85%.
3. •Generates churn_report.html with feature importance plot.