Data Charts and Moving Average

Choosing a data chart

A histogram displays one variable in equal width bins. Choose the bin count yourself. Bins include their lower edge and exclude their upper edge, except for the final bin which includes the maximum. Constant data produce one bin. Box plot quartiles use linear interpolation at the zero based index (n − 1)p in sorted data. Other quartile conventions can give different results.

Calculations and charts

Whiskers end at the most extreme observations within 1.5 IQR of the quartiles. Observations outside these fences appear as separate points and are not removed. For a scatter plot enter x and y on each line. The moving average uses complete trailing windows. Values 2, 4, 6, 8 with a window of 2 give 3, 5, 7. This is not a forecasting model.

Limits

Enter up to 2000 observations. Decimal commas are accepted in values separated by spaces when no decimal point occurs. Absolute values must not exceed 1e100. At most 50 bins are allowed. Units are not converted. Very close large values have limited numerical precision. The table below the chart contains all results. CSV saves the table and SVG saves the chart.

Ask different questions of the same list

Use the list 1, 2, 3, 4, 9, 10 in a histogram with two bins. Bin width is (10 − 1) / 2 = 4.5. The first bin includes its lower boundary 1 but excludes upper boundary 5.5 and contains four observations. The second bin from 5.5 to 10 contains two observations. Changing bin count can change appearance, but total frequency must still equal the observation count.

A moving average of three observations in the same order gives 2, 3, approximately 5.333333 and 7.666667. The first result appears at the third observation because earlier windows are incomplete. Reordering changes this result. Order is irrelevant for a histogram but matters for a moving average.

Check that all observations use the same unit before plotting. Each pair in a scatterplot row must belong to the same observation. Increasing together does not establish causation. Separate points on a boxplot are not automatically data errors. Before making decisions, inspect the table below the chart and how the records were collected.