Robust Average

A Python package that intelligently selects the most robust average (mean, median, or mode) for price analysis based on outlier and skewness detection.

Installation

pip install robust-average

Quick Start

from robust_average import robust_average

# Example with clean data
prices = [97.87, 109.99, 129.99, 89.99, 119.99]
result = robust_average(prices)
print(f"Selected average: {result['value']} (method: {result['method']})")
# Output: Selected average: 109.99 (method: mean)

# Example with outliers
prices_with_outlier = [97.87, 109.99, 129.99, 89.99, 119.99, 500.00]
result = robust_average(prices_with_outlier)
print(f"Selected average: {result['value']} (method: {result['method']})")
# Output: Selected average: 109.99 (method: median)

Features

• Automatic Method Selection: Intelligently chooses between mean, median, or mode
• Outlier Detection: Uses IQR method to identify and handle outliers
• Skewness Analysis: Measures data distribution asymmetry
• Transparent Results: Returns the method used and supporting statistics
• Business Ready: Ensures accurate price reporting for contracts and compliance

How It Works

The function uses a systematic approach to select the most appropriate average:

1. Outlier Detection (IQR Method):

   • Calculates Q1 (25th percentile) and Q3 (75th percentile)
   • Defines outliers as values outside [Q1 - 1.5×IQR, Q3 + 1.5×IQR]

2. Skewness Analysis:

   • Calculates skewness coefficient using scipy.stats.skew()
   • Values close to 0 indicate symmetric distribution

3. Decision Criteria:

   • Use MEAN if: No outliers AND |skewness| < 0.5
   • Use MEDIAN if: Outliers present OR |skewness| ≥ 0.5
   • Use MODE if: A single value appears in >50% of the dataset

🛠️ Process Flow

API Reference

robust_average(prices, return_all_stats=False)

Parameters:

• prices (list or pd.Series): List or Series of numeric prices
• return_all_stats (bool): If True, returns all computed statistics

Returns:

{
    'value': selected_average_value,
    'method': 'mean' | 'median' | 'mode',
    'mean': mean_value,
    'median': median_value,
    'mode': mode_value_or_None,
    'std': standard_deviation,
    'skew': skewness,
    'outliers': list_of_outlier_values,
    'count': number_of_prices
}

Use Cases

• Price Analysis: Ensure accurate average prices for reporting
• Contract Negotiations: Use defensible statistics in pricing discussions
• Compliance Reporting: Meet regulatory requirements with robust averages
• Data Quality: Automatically handle messy, real-world price data

Requirements

• Python >= 3.8
• numpy >= 1.20.0
• pandas >= 1.3.0
• scipy >= 1.7.0

License

MIT License

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.