> ## Documentation Index
> Fetch the complete documentation index at: https://podonos.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Analysis

> Understanding your evaluation results through various charts and metrics

## Analysis Overview

After completing your evaluation, you can analyze the results through various charts and metrics. The analysis dashboard provides comprehensive insights into your evaluation data. Want to see these analysis features in action? Check out our [sample evaluation analysis](https://workspace.podonos.com/Podonos/evaluations/fe4f1336-3683-45cf-a0fb-f6cba1a8db69?tab=analysis) to explore a real-world example of how these charts and metrics come together.

## Mean Score Analysis

### Single Evaluation Charts

<Steps>
  <Step title="Mean Scores by Model">
    View the average scores for each model in your evaluation through a bar chart visualization.
    ![single\_mean\_by\_model](https://static-public.podonos.com/sdk/analysis/v1/single_mean_by_model.png)
    This chart helps you quickly compare the performance across different models.
  </Step>

  <Step title="Mean Scores by Tag">
    Analyze average scores grouped by tags you've assigned.
    ![single\_mean\_by\_tag](https://static-public.podonos.com/sdk/analysis/v1/single_mean_by_tag.png)
    This visualization helps identify patterns across different categories or attributes.
  </Step>
</Steps>

### Double Evaluation Charts

<Steps>
  <Step title="Comparative Mean Scores by Model">
    For comparative evaluations, see which model was preferred in direct comparisons.
    ![double\_mean\_by\_model](https://static-public.podonos.com/sdk/analysis/v1/double_mean_by_model.png)
    This chart shows the preference distribution between Model A and Model B.
  </Step>

  <Step title="Comparative Mean Scores by Tag">
    Compare performances across different tags in paired evaluations.
    ![double\_mean\_by\_tag](https://static-public.podonos.com/sdk/analysis/v1/double_mean_by_tag.png)
  </Step>
</Steps>

## Detailed Analysis

### Response Distribution

<Steps>
  <Step title="Answer Frequency by Model">
    See how responses are distributed across different models.
    ![answer\_frequency\_by\_model](https://static-public.podonos.com/sdk/analysis/v1/common_answer_frequency_by_model.png)
    This horizontal bar chart shows the frequency of each response option per model.
  </Step>

  <Step title="Answer Frequency by Tag">
    Analyze response patterns based on tags.
    ![answer\_frequency\_by\_tag](https://static-public.podonos.com/sdk/analysis/v1/common_answer_frequency_by_tag.png)
  </Step>

  <Step title="Per-Query Response Distribution">
    View detailed response distributions for individual queries.
    ![answer\_frequency\_per\_file](https://static-public.podonos.com/sdk/analysis/v1/common_answer_frequency_per_file.png)
    This helps identify specific queries that received particular response patterns.
  </Step>
</Steps>

### Query-Level Analysis

<Steps>
  <Step title="Query Summary">
    View a comprehensive list of all quries with their:

    * Associated models
    * Tags
    * Mean scores
      ![file\_summary](https://static-public.podonos.com/sdk/analysis/v1/common_file_summary.png)
  </Step>

  <Step title="Query Details">
    Access detailed information about each evaluated file:

    * Individual responses
    * Evaluator demographics (nationality, gender)
    * Associated script (if available)
      ![file\_detail](https://static-public.podonos.com/sdk/analysis/v1/common_file_detail.png)
  </Step>

  <Step title="Interactive Scatter Plot">
    Explore all query scores in an interactive scatter plot:

    * Filter by model
    * Customize chart views
    * Identify patterns and outliers
      ![mean\_scatterchart](https://static-public.podonos.com/sdk/analysis/v1/common_mean_scatterchart.png)
  </Step>
</Steps>

<Note>
  **Pro Tip**: Make the most of your analysis by utilizing tags effectively. Tags allow you to slice and dice your data in various ways, providing deeper insights into specific aspects of your evaluation. Consider adding tags for characteristics like:

  * Speaker demographics (gender, age group)
  * Audio characteristics (noisy, clean)
  * Content type (question, statement)
  * Any other relevant categorization
</Note>