How to Calculate Top N Percentage of Products in Python

Introduction

When analyzing product data, it's often useful to identify the top-performing items based on a specific metric. This could be sales volume, revenue, profit margin, or any other key performance indicator. Calculating the top N percentage allows you to focus on the most significant products while ignoring the rest.

In this guide, we'll cover:

• The mathematical approach to calculating top percentages
• Python implementation using pandas and numpy
• Practical examples with real-world data
• Visualization techniques to present the results

Methodology

The basic approach involves:

1. Sorting the products by the relevant metric in descending order
2. Calculating the cumulative sum of the metric values
3. Identifying the point where the cumulative sum reaches the desired percentage
4. Selecting all products up to that point

This method ensures you capture the most significant products while maintaining the integrity of the percentage calculation.

Python Implementation

Here's a complete Python implementation using pandas and numpy:

import pandas as pd
import numpy as np

def calculate_top_percentage(df, metric_column, percentage):
    """
    Calculate top N percentage of products based on a metric.

    Parameters:
    df (DataFrame): Input data
    metric_column (str): Column name containing the metric values
    percentage (float): Desired percentage (0-100)

    Returns:
    DataFrame: Top N percentage of products
    """
    # Sort by metric in descending order
    sorted_df = df.sort_values(by=metric_column, ascending=False)

    # Calculate cumulative sum
    sorted_df['cumulative'] = sorted_df[metric_column].cumsum()

    # Calculate total sum
    total_sum = sorted_df[metric_column].sum()

    # Find the cutoff point
    cutoff = (percentage / 100) * total_sum
    top_products = sorted_df[sorted_df['cumulative'] <= cutoff]

    return top_products.drop(columns=['cumulative'])

# Example usage
data = {
    'product_id': [1, 2, 3, 4, 5, 6, 7, 8, 9, 10],
    'product_name': ['A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J'],
    'sales': [100, 200, 150, 300, 250, 180, 220, 170, 190, 210]
}

df = pd.DataFrame(data)
top_50_percent = calculate_top_percentage(df, 'sales', 50)
print(top_50_percent)

The function sorts the products by sales in descending order, calculates the cumulative sum, and then selects products until the cumulative sum reaches 50% of the total sales.

Example

Let's work through an example with the following product sales data:

Calculating the top 50% of products by sales:

1. Total sales = 100 + 200 + 150 + 300 + 250 + 180 + 220 + 170 + 190 + 210 = 1800
2. 50% of total sales = 900
3. Sort products by sales (descending): D (300), B (200), E (250), G (220), J (210), etc.
4. Cumulative sum reaches 900 after including D (300), B (200), E (250), G (220), J (210)

The top 50% of products by sales are D, B, E, G, and J.

Visualization

Visualizing the top N percentage of products can provide valuable insights. Here's how to create a bar chart showing the top products:

import matplotlib.pyplot as plt

def visualize_top_products(df, metric_column, percentage):
    top_products = calculate_top_percentage(df, metric_column, percentage)

    plt.figure(figsize=(10, 6))
    plt.bar(top_products['product_name'], top_products[metric_column])
    plt.title(f'Top {percentage}% Products by {metric_column.capitalize()}')
    plt.xlabel('Product')
    plt.ylabel(metric_column.capitalize())
    plt.xticks(rotation=45)
    plt.tight_layout()
    plt.show()

# Example usage
visualize_top_products(df, 'sales', 50)

This will generate a bar chart showing the top 50% of products by sales, making it easy to compare their performance.