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

# stats_utils

> Statistics calculation and summary generation utilities

# browseruse\_bench.utils.stats\_utils

Statistics-related utility functions.

## Import

```python theme={null}
from browseruse_bench.utils import (
    calculate_metric_stats,
    calculate_all_metrics_stats,
    filter_tasks_by_label,
    generate_evaluation_summary,
)
```

***

## calculate\_metric\_stats

Calculate statistics for a specific metric.

```python theme={null}
def calculate_metric_stats(
    tasks: List[Dict[str, Any]],
    metric: str,
    path: str = "evaluation_details"
) -> Dict[str, float]
```

<ParamField path="tasks" type="list of dict" required>
  List of task results
</ParamField>

<ParamField path="metric" type="str" required>
  Metric name, e.g., `ttft_ms`, `end_to_end_ms`, `steps`
</ParamField>

<ParamField path="path" type="str" default="evaluation_details">
  Path to metric in task dictionary
</ParamField>

<ResponseField name="return" type="dict">
  Statistics dictionary containing `count`, `mean`, `min`, `max`, `median`
</ResponseField>

***

## calculate\_all\_metrics\_stats

Calculate statistics for multiple metrics.

```python theme={null}
def calculate_all_metrics_stats(
    tasks: List[Dict[str, Any]],
    metrics: Optional[List[str]] = None,
    path: str = "evaluation_details"
) -> Dict[str, Dict[str, float]]
```

<ParamField path="metrics" type="list of str" default="see below">
  List of metric names
</ParamField>

### Return Structure

```python theme={null}
{
    "ttft_ms": {"count": 10, "mean": 1234.5, ...},
    "end_to_end_ms": {"count": 10, "mean": 5678.9, ...},
    "steps": {"count": 10, "mean": 3.2, ...},
    "usage": {
        "total_tokens": {"count": 10, "mean": 1500, ...},
        "total_cost": {"count": 10, "mean": 0.05, ...},
        ...
    }
}
```

***

## filter\_tasks\_by\_label

Filter tasks by label.

```python theme={null}
def filter_tasks_by_label(
    tasks: List[Dict[str, Any]],
    key: str = "predicted_label",
    val: int = 1
) -> List[Dict[str, Any]]
```

<ParamField path="tasks" type="list of dict" required>
  List of task results
</ParamField>

<ParamField path="key" type="str" default="predicted_label">
  Label key name
</ParamField>

<ParamField path="val" type="int" default="1">
  Label value (1 = success, 0 = failure)
</ParamField>

***

## generate\_evaluation\_summary

Generate evaluation summary.

```python theme={null}
def generate_evaluation_summary(
    results: List[Dict[str, Any]],
    total: int,
    metrics: Optional[List[str]] = None
) -> Dict[str, Any]
```

<ParamField path="results" type="list of dict" required>
  List of evaluation results
</ParamField>

<ParamField path="total" type="int" required>
  Total number of tasks
</ParamField>

<ParamField path="metrics" type="list of str" default="see below">
  List of metrics to calculate
</ParamField>

### Return Structure

```python theme={null}
{
    "overall_statistics": {
        "total_tasks": 100,
        "evaluated_tasks": 95,
        "successful_tasks": 70,
        "failed_tasks": 25,
        "success_rate": 73.68,
        "failure_rate": 26.32
    },
    "metrics_statistics": {...},
    "successful_tasks_metrics": {...},
    "failed_tasks_metrics": {...},
    "failure_category_statistics": {...},
    "task_list": {
        "successful_task_ids": [...],
        "failed_task_ids": [...]
    }
}
```