Binomial Confidence Intervals Visualizer

A Streamlit app for visualizing confidence intervals for TPR (True Positive Rate) and TNR (True Negative Rate) across different sample sizes.

Features

• Interactive sliders to adjust:

  • TPR and TNR values
  • Prevalence
  • Sample size range
  • Confidence level (via alpha)
  • Number of points to calculate

• Three-panel visualization:

  • TPR confidence intervals
  • TNR confidence intervals
  • Width of confidence intervals

Installation

1. Clone this repository
2. Install the required packages:

pip install -r requirements.txt

Usage

Run the Streamlit app:

streamlit run app.py

Then access the app in your web browser (typically at http://localhost:8501).

Parameters

• TPR: True Positive Rate (Sensitivity) - The proportion of actual positives correctly identified
• TNR: True Negative Rate (Specificity) - The proportion of actual negatives correctly identified
• Prevalence: The proportion of positive cases in the population
• Sample Size: Min and max total sample size to visualize
• Significance Level (α): Determines the confidence level (1-α) * 100%
• Number of Sample Points: Number of points to calculate along the sample size range