Analytics (Beta)
Analytics lets you ask questions of your history across many devices at once: the average signal strength per site over the last 90 days, water use by hour of day, which devices run hottest. You build a visualization, save it to your library, and, if you like, place it on any of your dashboards as a tile.
This is a beta feature. It appears only when your plan includes it. Chart types and options will keep growing, and a few details may change as it matures. If you do not see Analytics (Beta) in the sidebar, your plan does not include it yet; ask your provider.
Where to find it: Sidebar -> Analytics (Beta) -> Visualizations.
Dashboards or Analytics?
Dashboards show what is happening now: live readings, gauges, alerts, and short time ranges of up to 30 days. Analytics looks back. It aggregates readings across devices over days, weeks, or months, and lets you group and compare them. A visualization built in Analytics can then sit on a dashboard next to your live tiles, so both views share one screen.
Who can use it
- Anyone whose role can see the Analytics section can open the library, view visualizations, and run them.
- Building, editing, and deleting visualizations needs the Manage analytics permission. Administrators have it by default. An Operator or Viewer can be given it by an administrator in Account settings under Role permissions.
The library
The Visualizations page lists everything saved in your workspace: name, description, chart type, when it last ran, and when it was last updated. Click a row to open it.
New visualization starts a new one. If the button reads Limit reached, your plan's cap on saved visualizations is full: delete one you no longer use, or ask your provider about a higher limit.
Building a visualization
Step 1: choose a chart type
Pick the kind of chart first. The type decides what the editor asks you for: whether it groups by a category, whether it has a time axis, and how many values it takes.
| Type | Best for | What it needs |
|---|---|---|
| Bar | Comparing or ranking a measure across sites or devices | One or more values; optional Group by; can be stacked or horizontal |
| Pie | The share of a positive total across a category | One value and a Group by (positive values only) |
| Scatter | The relationship between two measures, one point per site or device | Exactly two values and a Group by |
| Line | A trend over time | One or more values and a time grain; optional Group by for one series per site or device |
| Area | A filled trend over time, optionally stacked | Same as Line |
| Heatmap | Intensity by hour of day across weekdays or days | One value; the axes are chosen on the Settings tab |
| Table | The numbers behind any binding, as rows | Any binding |
A single current reading, a gauge, or a map of live positions belongs on a dashboard, where the Value, Gauge, and Map widgets show what is happening now. Analytics is for what happened over time.
Step 2: the Data tab
| Field | What it does |
|---|---|
| Devices | Which devices feed the chart: All devices, one device profile, a site, a network, or specific devices. When every selected device shares one profile, that profile's own readings (temperature, flow, and so on) become available as values. A mixed selection charts the universal signals every device reports (signal strength, signal-to-noise, signal quality, and gateways per reading) plus any semantic fields. |
| Values | What to measure and how to summarize it: average, minimum, maximum, sum, count, latest, or change (last minus first, for meters that count up). Most types take up to six values; Scatter exactly two, Pie and Heatmap one. A profile's formula and mapping fields are listed alongside its readings, each marked with its kind; see Formula and mapping fields in Analytics. A Semantic fields group lists readings that have been marked as the same kind across device profiles, such as Temperature (°F) or Humidity (%): pick one and every device whose profile has that reading contributes its own field, whatever the device selection. Devices without it contribute nothing to that value and stay in the chart's other values. A count of such a value counts the readings that carry it. |
| Group by | Splits the chart by site, device, hour of day, or day of week. On a Line or Area chart each group becomes a series; on Bar, Pie, Table, and Scatter each group becomes a bar, slice, row, or point. |
| Period | How far back to look: the last 7, 30, 90, 180, or 365 days, ending now. Viewers can be allowed to change it; see the Filters tab. |
| Time grain | For Line and Area: the bucket size, hourly, daily, weekly, or monthly. Buckets follow your time zone. |
| Description | Optional. Shown in the library so others know what the chart answers. |
The preview re-runs as you change the data. The caption under it tells you how many rows came back, which level of history served them (readings, hourly summaries, or daily summaries), how fresh hourly summaries are, how long the query took, and which time zone the times are in.
Values show with the decimals set on the field in the device profile, in the chart, in tooltips, and in the downloaded file. Signals use fixed decimals: whole numbers for signal quality and signal strength, one decimal for signal-to-noise. A field with no decimals set shows whole numbers from 100 up, one decimal from 10 up, and two below; in the downloaded file it keeps its precision, to six significant digits. A count is always a whole number.
Formula and mapping fields in Analytics
A formula field, for example a temperature in °F derived from °C, and a mapping field that yields a number chart like any other value, over any period and at any time grain. The value is computed when each reading arrives and stored with it, so Analytics shows exactly what the dashboard shows. A reading that only reports on change is carried forward into that computation, so a formula over it has a value on every reading.
A field's history starts at its first stored value. A formula or mapping field added to a profile, or changed, applies from the point chosen when it is saved; readings before that point show no value for it until its history is recomputed from the profile's fields. A reading that is calibrated per device charts with the calibration applied.
When a chart would be too big
A chart holds up to 100,000 rows (time buckets times groups): a year of hourly values for about ten devices, or a year of daily values for a few hundred. A Table holds 10,000 rows, since it shows every one. Rather than quietly cutting the data short, Analytics checks the size before running and refuses a query that would not fit, naming the size it would have been and what does fit, for example "This would be about 439,000 rows (8,784 hourly buckets x 50 groups); the most a chart can hold is 100,000. Daily fits (18,300 rows), or keep Hourly over the last 60 days (72,000 rows)."
Two more checks work the same way. A chart that would read too much history for one query, for example every device in a large fleet at hourly detail over a year, is refused with the size and the grain that fits. A heatmap of hours by day covers at most 92 days. The visualization cannot be saved until its query fits.
Step 3: the Filters tab
| Field | What it does |
|---|---|
| Filters | Numeric conditions on the signal values or on semantic fields, for example only readings with a signal strength above -100 dBm, or only readings with a Temperature (°F) above 80. When every selected device shares one profile, that profile's readings can be filtered too; the tab says so when it needs that profile choice on the Data tab. Filters read the detailed readings, which are kept for a limited time, so a filtered chart cannot reach beyond them, and a filtered chart over many devices and a long period may be refused as too big. |
| Viewer filters | Whether viewers may change the period the chart covers, on dashboard tiles and in Analytics, and which selector they see: a Date range (a from and a to day) or a Relative period (last 7 days, 30 days, 3, 6, or 12 months, this month, this year). Off unless you turn it on. The saved period is always the starting point. See Choosing the period yourself. |
When Viewer filters is on, a Preview as viewer switch above the preview shows you the control your viewers get, so you can try it before saving.
Step 4: the Settings tab
Look-and-feel, and what a click does. Changing these never re-queries the data.
| Setting | What it does |
|---|---|
| Series | A color and a label for each value. |
| Fixed lines | Reference lines and bands drawn over a Bar, Line, or Area chart: a target, a limit, a comfort band, each with a label and a color. |
| Drill-down | Whether viewers can click into the chart, the ordered steps a click takes, and whether viewers choose the next step themselves. See Drilling into a chart. |
| Data export | Whether viewers can download the rows behind the chart as a CSV file, in Analytics and on dashboard tiles. On by default. See Downloading the data. |
| Bars, Areas | Stack the values; draw bars horizontally; print the value on each bar, or the total on each stack. |
| Pie | Doughnut style, percentage labels, and how many slices to show before the smaller ones fold into "Other". |
| Points | Scatter: label each point with its name. |
| Axes, Colors, Cells | Heatmap: weekday by hour or day by hour; one hue from low to high, or your own ranges with a color and a label each (green for normal, red for out of bounds); print the value inside each cell when there is room. |
| Axes, Legend | Axis titles, the value axis, and the legend on Bar, Line, and Area charts. |
Step 5: save
Give the visualization a name and click Save. It appears in the library straight away and can be placed on dashboards from that moment.
Reading a visualization
- Times are yours. Buckets, hours of the day, and days of the week follow the time zone in your profile, and the caption under the chart says which one. A colleague in another time zone sees the same readings grouped by their own hours.
- How far back it can look. Detailed readings are kept for a limited time, hourly summaries for about a year, and daily summaries for years. Windows of up to two days at hourly detail use the readings; anything longer uses the summaries, which hold the same numbers and are refreshed every few minutes, and the caption under the chart says which. Filters need the readings, so a filtered chart cannot reach beyond them. A daily chart over more than 60 days is drawn on UTC days and says so. If a period reaches further back than is kept, the chart uses what is available and notes the earliest date it could use.
- Names, not IDs. Sites and devices show by name. Readings from devices with no site show as Unassigned.
- Large charts are drawn from a sample. A Line or Area chart with more points than the screen can show draws a chosen subset of its time buckets, keeping every peak and dip, and says "Drawn from 61,320 rows" under the chart. Dots come off dense lines. The download and the table still hold every row.
- Pie needs positive values. A measure that can be negative or zero, such as signal strength, shows a message pointing you to Bar or Table instead.
- Too many preview runs in a short burst, or Analytics is busy when many charts run at once, shows a short message and retries on its own. Give it a moment and continue.
Drilling into a chart
A chart that allows it can be clicked to look closer. Each click re-queries the data with a narrower window or a narrower group, so what you see is always real numbers, never a zoom of the picture:
- On a Line or Area chart, click a point to see that bucket at the next finer level: a day opens into its hours. On a chart drawn from a sample, the click covers the span up to the next drawn point, so every hour of a long chart stays reachable, and the path names the span. When the next step is a group, the click splits the window into one series per site or device instead, and clicking a series name in the legend narrows to that one.
- On a Bar or Pie chart, click a bar or a slice to see the devices behind that site.
The path you have taken shows above the chart, for example "Last 30 days › Aug 16 › 14:00". Click an earlier crumb to step back. The caption beside it says what the next click will do, or why a step is not available: hourly detail, for example, is only kept for a limited time, so a day too far back cannot open into hours.
Whoever built the visualization decides all of this on its Settings tab under Drill-down: whether clicks drill at all, the order of steps (a time level, then a group), and whether viewers pick the next step themselves. When that last option is on, a click offers the available steps and you choose.
Drill-down works the same way on a dashboard tile: click into the chart there, and the path shows inside the tile. Each viewer's drill is their own and resets when the page reloads, and when you change the chart's period (see below) the drill starts over from the new window.
Choosing the period yourself
When the author turned on Viewer filters, a calendar button sits at the top of the tile, and above the preview in Analytics. Click it to change the period the chart covers:
- Date range: pick a from and a to day. The window can be up to a year, and a heatmap of hours by day up to 92 days.
- Relative period: last 7 days, 30 days, 3, 6, or 12 months, this month, or this year.
A dot on the button shows the chart is not on its saved period, and the button's tooltip names the window. The choice is yours alone and is not saved: the chart returns to its saved period when the page reloads, and other viewers see their own choices. A window that is too big for the chart is refused with what fits, the same way as in the editor.
Downloading the data
Every visualization can hand you the rows behind it as a CSV file that opens in Excel and other spreadsheet apps:
- In Analytics, click the download icon above the preview.
- On a dashboard, click the download icon in the tile's header.
The file holds exactly what is on screen: the same columns (time, the group, each value with its unit), sites and devices by name, and, when you have drilled into the chart or changed its period, only the rows at that level and window. Every row is included, even when the chart itself is drawn from a sample; a file holds up to 100,000 rows, the chart's own cap.
Whoever built the visualization decides whether the download is offered, on its Settings tab under Data export. It is on unless they turned it off.
Putting a visualization on a dashboard
- Open the dashboard and turn on Edit.
- Click + Add visualization, next to + Add widget.
- Pick a visualization from your workspace's library. The tile drops into the next open slot.
- Open the tile's settings (gear) to change which visualization it shows, how often it refreshes (every minute up to every hour), and whether the period prints under the chart.
The tile shows the visualization at its saved period, for example the last 90 days, and refreshes on its own schedule. It does not follow the dashboard's time-range picker, but when the author allows it, each viewer can change the tile's period from the calendar button on the tile; see Choosing the period yourself. Tiles are not included when a dashboard is shared by public link.
Editing and deleting
- Open a visualization from the library to change it. Every dashboard tile that shows it picks up the change on its next refresh.
- Delete from the library. If dashboard tiles still use the visualization, the confirmation tells you how many; those tiles then read "Visualization removed" until you point them at another visualization or remove them.
Frequently asked
Why does my Pie chart show nothing?
A Pie shows the share of a positive total. Signal strength is negative, so its slices cannot be drawn; use a Bar or a Table for that measure, or pick a positive measure such as a count.
Why does a chart say a selected device is not available?
A device picked on the Data tab is no longer active in your workspace. The chart leaves it out and notes how many devices are not shown; when none of the selected devices can be shown, it says Device not available. The device appears as Unavailable device in the device list, where you can remove it or pick another. See When a device is not available.
My hour-of-day numbers differ from a colleague's. Why?
Hours follow each viewer's own time zone, set in your profile. Both of you are looking at the same readings, grouped by your own local hours.
Can a visualization follow the dashboard's time range?
No. A tile shows its saved period so that a 90-day trend stays a 90-day trend wherever it is placed. What a tile can do, when its author turned on Viewer filters, is let you change its own period from the calendar button on the tile. See Choosing the period yourself.
Can I click into a chart to see more detail?
Yes, when its author left drill-down on, in Analytics and on dashboard tiles alike. See Drilling into a chart.
Can I get the numbers behind a chart?
Yes, as a CSV file, from the download icon in Analytics or on the tile, unless the author turned data export off. See Downloading the data.