Understanding Mode and Median in Data Analysis

Mode for Continuous Data

Continuous data presents a unique challenge. Because continuous variables (like annual insurance premiums or precise weights) can be measured to infinite decimal places, it is highly unlikely that any two exact values will repeat. If every value in your dataset is unique, a traditional mode does not exist.

To solve this, we use grouped continuous data. Instead of looking for a single repeating number, we group the data into intervals called bins.

Defining Bins and the Modal Class

By defining bins, we can categorize data points into ranges. For example, instead of looking for the exact premium of $$16,884.924$, we might create a bin for premiums between $$15,000$ and $$20,000$.

The bin that contains the highest frequency of data points is called the modal class. There are two ways to estimate the mode from this class:

1. Crude Mode: A simple approximation where we take the middle value of the modal class.
2. True Mode: A more precise calculation for grouped data that accounts for the distribution of data in the bins surrounding the modal class. For instance, if there is significantly more data in the bin to the right of the modal class than to the left, the "true" mode will be shifted toward the right.

The Median: The Middle Value

The median is another powerful measure of typical behavior. Unlike the mode, which looks for frequency, the median identifies the middle value of a dataset once it has been ordered from smallest to largest.

Calculating the Median

To find the median, you must first sort your data. The method of finding the center depends on the number of data points ($n$):

• Odd number of points: The median is the exact middle value.
• Even number of points: The median is the average of the two middle values.

For example, if you have a sorted list of insurance charges, you would count to the center to find the representative value.

Important Limitation: Because the median relies on a specific order, it is not appropriate to use for Nominal Data (like favorite colors), as you cannot "order" colors from smallest to largest.

Resilience to Outliers

One of the greatest strengths of the median is that it is the least sensitive to extreme values or outliers.

In a dataset where most values are small but one value is extremely large (an outlier), the mean will be pulled upward, potentially giving a misleading sense of "typical" behavior. The median, however, remains anchored to the center of the distribution, providing a more stable reflection of the majority of the data.