analyze-results

To analyze ML experiment results across multiple seeds and runs, the skill computes summary statistics like mean and standard deviation, builds comparison tables with delta versus baseline, and generates narrative insights for decision making.

Generating comparison tables with delta versus baseline is handled by structuring your experiment results directory as input, which the skill processes to produce comparative tables of models, hyperparameters, and metrics automatically.

The best way to compute reproducibility checks and statistics for ML experiments is to apply structured analysis to your results dataset, producing an interpretable narrative alongside a reproducible task outline and computed standard deviations.

Yes, you can analyze experiment results with varying models and hyperparameters, as the skill is specifically designed to process datasets with multiple runs, varying models, and different hyperparameter configurations to derive actionable insights.

Turning machine learning experiment results into clear insights involves computing statistics and building tables, which the skill uses to automatically generate narrative findings and recommended next steps for your experiments.

Analyzing experiment results with this skill requires no specific dependencies or components, allowing you to directly process your results directory to generate structured summaries and insights without additional environment setup.