The Time Based Chart

Creating one#

Create it with the constructor for the interval you want — Hourly, Daily, Monthly, and so on. That fixes the interval permanently, and the chart then accepts only measures at that interval.

Drop measures onto the container that accepts them and the chart type builds the axis, stack, and series structure for you. See How Infohub Charts Work.

Series types#

Each series can be drawn differently, and the choice carries meaning:

The distinction that matters most: lines for rates, bars for quantities. A line implies the value existed continuously between the points, which is true for a temperature and false for a daily total. A bar implies a discrete period, which is right for the total and wrong for the temperature.

Step is underused. A setpoint or a tariff rate does not slide smoothly from one value to the next — it holds and then jumps. Drawing it as a line invents a transition that did not happen.

Axes#

A chart can carry more than one axis, which is how you plot quantities of different magnitude or unit together.

Guidance that keeps charts honest:

Include zero on bar charts. A bar chart whose axis starts at 4,000 exaggerates differences dramatically. Bars encode magnitude by length, and truncating the axis breaks that.

Truncating is legitimate on line charts. A temperature that operates between 85 and 105 °C is unreadable on a 0–105 axis. Line charts encode change rather than magnitude, so a tight range is informative.

Fix the range when comparing. Two charts side by side with independent auto-ranged axes will be read as comparable and are not. If people will compare them, fix both to the same range.

Two axes maximum. Three or more and nobody can tell which series belongs to which. If you need three units, you need more than one chart.

Limits on a chart#

Limits show the thresholds a measure is judged against — target, upper, lower — drawn on the chart so a reader can see at a glance whether values are acceptable.

Limits belong to the measure, not the chart. To show one, the chart must already carry that measure as a series, and then the limit is bound onto it. A limit slot only accepts a limit belonging to the measure on its parent item, so the order is: add the series, then bind the limit.

This is worth doing on any chart people are expected to act on. A trend with no reference line requires the reader to remember what good looks like. See Measure Limits and Thresholds.

Period comparison#

To compare against a previous period, use the Time Based Comparison Chart rather than building the comparison by hand.

The alternative — a calculated measure reaching back twelve periods, plotted alongside the current one — also works and is sometimes what you want:

Measure_1043[-12].Value

The comparison chart type is less work and keeps the intent obvious. Reach for the calculated approach when the comparison is itself a number people want, such as a variance percentage, rather than just a second line.

How many series?#

Fewer than you want to add.

Three or four series is readable. Eight is a tangle that nobody interprets, and the usual reason it happens is trying to answer several questions on one chart.

Practical approach: one question per chart. "How did the boilers compare" is one chart with three lines. "How did Boiler 1 perform" is a different chart with flow, pressure, and temperature. Trying to do both produces nine series and no insight.

Where you genuinely need many series, a heat map often shows a large set more clearly than a line chart can.

Regression and trend lines#

Any measure on an axis can carry a regression, fitting a trend to the plotted series. The Scatter chart supports the same thing, where it is often the more natural home — a regression through a scatter of two variables is the classic use.

One property of it changes how the result should be read:

That makes it excellent for exploration and poor as a reported number. If a trend or a projection needs to be reported, agreed, or alarmed on, compute it as a calculated measure instead — the Regression library exists for exactly this, with LineFitSlope, LineFitIntercept and LineFitPredicted. A measure has a defined period, is stored, and can carry a limit. See Built-in Function Reference.

Use the chart regression to see whether a relationship is there. Use a measure once you want to act on it.

Interpolation and gaps#

Missing values are a real editorial decision, not a display detail.

A chart can join across a gap or break the line. Breaking is more honest — it shows the reader that data was absent rather than implying a smooth transition through it. Joining across a two-day outage draws a straight line that looks like real, gradual movement.

Where gaps are frequent and meaningful, consider whether the underlying measure should be returning null rather than zero, since a chart can only display the distinction if the measure preserves it. See Calculated Measures.

Units and formats come from the measure#

Axis labels, tooltips, and legends use the measure's unit and format string. Set those on the measure and every chart agrees; override them per chart and the same quantity will be displayed differently in different places.

This is the single easiest way to keep a large dashboard set consistent — get the measure right once. See Format Strings and Units of Measure.

Choosing the interval#

The interval should match the decision the chart supports.

A monthly chart is the wrong tool for diagnosing this morning, and a minutely chart is useless for a board pack. Building one chart and hoping it serves both is the most common cause of dashboards nobody uses.

Diagnosing#

See also#

• How Infohub Charts Work
• Chart Type Catalogue
• Measure Limits and Thresholds
• Dashboard Layout Principles