Calculate relative frequency, cumulative frequency, percentages, and Shannon entropy from ungrouped data. Includes bar chart, ogive curve, and frequency distribution table.
The relative frequency calculator converts raw counts into proportions, percentages, and cumulative frequencies for discrete data. Relative frequency (count / total) is the observed share of each value and a practical way to compare uneven categories.
This tool builds a frequency table with absolute frequency, relative frequency, percentage, cumulative frequency, and cumulative percentage. It also shows the mode, Shannon entropy, a bar chart, and an ogive so you can read both the shape and the spread of the data.
Enter discrete values and use the table to see which outcomes dominate, how quickly the totals accumulate, and whether the distribution is balanced or concentrated in just a few values.
Use this calculator when you need to turn counts into proportions, compare categories with different totals, or inspect whether one value dominates the dataset. The cumulative columns make it easy to read running totals, while entropy gives a compact measure of how even the distribution is.
It is useful for survey answers, dice rolls, classroom data, quality checks, and any small discrete dataset where a table and chart are easier to read than a raw list.
Relative frequency = fᵢ / n where fᵢ is the frequency and n is the total. Cumulative frequency = sum of all frequencies up to and including current value. Shannon entropy H = −Σ pᵢ log₂(pᵢ). Normalized entropy = H / log₂(k) where k = number of unique values.
Result: Mode = 3 (freq 5), Entropy = 2.15 bits
Value 1: 1/15 = 6.67%. Value 2: 2/15 = 13.33%. Value 3: 5/15 = 33.33% (mode). Value 4: 4/15 = 26.67%. Value 5: 3/15 = 20.00%. Shannon entropy = 2.15 bits out of max 2.32 bits, giving 92.6% normalized entropy — a fairly even distribution.
Relative frequency is the count for one value divided by the total number of observations. It is the simplest normalized description of a dataset, so the same table can be used even when sample sizes differ.
Cumulative frequency tells you how many observations are at or below a value. Cumulative relative frequency shows the same idea as a fraction or percentage, which makes percentile-style interpretation straightforward.
Shannon entropy is highest when the values are evenly distributed and lowest when one value dominates. For a discrete dataset, it gives a quick way to compare concentration without having to inspect every row of the table.
Last updated:
Relative frequency is the proportion of observations that have a specific value: frequency / total count. It ranges from 0 (value never appears) to 1 (all observations are this value). Multiplied by 100, it gives the percentage. Relative frequencies always sum to 1.0 (or 100%) across all values.
Frequency (absolute frequency) is the raw count of how many times a value appears. Relative frequency is that count divided by the total number of observations. Frequency depends on sample size; relative frequency is normalized to [0,1], making it comparable across different-sized datasets.
Cumulative relative frequency at value x is the sum of all relative frequencies for values ≤ x. It tells you what proportion of the data falls at or below x. Starting at 0 and ending at 1.0, it creates the empirical cumulative distribution function (ECDF) — also called the ogive when plotted.
Shannon entropy H = −Σ pᵢ log₂(pᵢ) measures the "uncertainty" or "information content" of a distribution. If one value dominates (low entropy), there's little surprise. If all values are equally likely (max entropy = log₂k), each observation is maximally informative. It's measured in bits (log₂) or nats (ln).
Relative frequency is an empirical (observed) measure from data. Probability is a theoretical concept. The law of large numbers guarantees that relative frequency converges to true probability as sample size grows. For small samples, relative frequency is a rough estimate; for large samples, it's a reliable estimate of probability.
Normalized entropy (H / Hmax) ranges from 0 to 1. Values near 1 mean the distribution is close to uniform (all values equally frequent). Values near 0 mean one value dominates. For a Likert scale survey: 95%+ normalized entropy suggests no clear preference; below 70% suggests strong preference for certain responses.