Histograms and Frequency Polygons
Histograms and frequency polygons are tools for visualizing how data is distributed. They let you quickly see the shape, center, and spread of a dataset, and they make it much easier to compare groups side by side.
Time series graphs serve a different purpose: they track how data changes over time, revealing trends, seasonal cycles, and unusual events.
Histograms
A histogram is a bar chart for continuous, numerical data. Unlike a regular bar chart (which is for categorical data), a histogram groups values into bins of equal width, and the bars touch each other because the data is continuous with no natural gaps.
Here's what each part does:
- Horizontal axis (x-axis): Shows the data values divided into equal-width intervals, or bins (e.g., age groups 0–9, 10–19, 20–29).
- Vertical axis (y-axis): Shows the frequency (count) or relative frequency (proportion/percentage) of observations falling in each bin.
- Bars: Each bar's height corresponds to how many data values fall in that bin. The bars have no gaps between them because the bins cover a continuous range.
Once you've built a histogram, you can read several things from it:
- Shape: Is the distribution symmetric, skewed left, skewed right, or bimodal (two peaks)?
- Center: The tallest bar indicates the mode, and you can estimate where the middle of the data falls.
- Spread: How wide the histogram stretches tells you about variability.
- Outliers: Isolated bars far from the main cluster suggest unusual values.
One thing to watch: changing the bin width can change how the histogram looks. Too few bins and you lose detail; too many bins and the graph gets noisy. There's no single "correct" number of bins, but your choice should reveal the overall pattern without distorting it.
Frequency Polygons
A frequency polygon is a line graph built from histogram data. Instead of bars, you plot a point at the midpoint of each bin (at the height of that bin's frequency), then connect those points with straight line segments.
How to construct a frequency polygon
- Set up bins with equal widths, just as you would for a histogram.
- Calculate the frequency (or relative frequency) for each bin.
- Find the midpoint of each bin. For a bin spanning 10–19, the midpoint is .
- Plot a point at each midpoint at the corresponding frequency.
- Connect the points with line segments.
- Close the polygon by extending line segments down to the x-axis at the midpoints of the (empty) bins on either end of the distribution. This "anchors" the polygon to zero on both sides.
Why use a frequency polygon instead of a histogram?
The main advantage is comparison. If you want to compare two or more distributions on the same graph (say, test scores for two different class sections), overlapping histograms get messy fast. Frequency polygons let you overlay multiple distributions cleanly, each as its own line, so differences in shape, center, and spread are easy to spot.
For example, you could plot male and female height distributions on the same axes, or compare income distributions across different age groups.
Time Series Graphs
A time series graph plots data values against time. The x-axis represents equally spaced time intervals (days, months, quarters, years), and the y-axis represents the measured variable (stock price, temperature, unemployment rate). Data points are connected by line segments to emphasize how the variable changes over time.
How to construct a time series graph
- Choose consistent time intervals for the x-axis (e.g., monthly from January 2020 to December 2024).
- Plot each data value on the y-axis at its corresponding time point.
- Connect consecutive data points with line segments.
What to look for
- Long-term trend: Is the variable generally increasing, decreasing, or staying flat over the full time span?
- Seasonal or cyclical patterns: Does the variable rise and fall in a repeating cycle? Retail sales, for instance, typically spike in December every year.
- Unusual observations: A sudden spike or drop that breaks the pattern may signal an external event (a policy change, a natural disaster, a market crash).
Common applications
- Economics: Tracking GDP growth, unemployment rates, or inflation over decades.
- Weather/climate: Plotting monthly average temperatures or annual precipitation.
- Business: Comparing quarterly revenue across multiple product lines on the same graph.
- Demographics: Visualizing population growth or birth rate changes over time.
Because time series graphs show data in chronological order, they're also the starting point for forecasting: you identify the trend and seasonal components, then project them forward.
Data Interpretation and Descriptive Statistics
Graphs are only useful if you can extract meaning from them. When interpreting any of these visualizations, keep a few things in mind:
- Context matters. A histogram of exam scores means something different depending on whether the exam was curved, how many students took it, and what the passing threshold is.
- Combine graphical and numerical summaries. A histogram shows you the shape; measures like the mean, median, and standard deviation give you precise numbers. Use both.
- Watch for misleading scales. A y-axis that doesn't start at zero, or bins of unequal width, can distort your perception of the data.
- Consider sample size. Small datasets can produce histograms that look irregular even when the underlying population is smooth. Larger samples give more reliable pictures.