Back to ConstantGraph homepage
Time Series

Time Series

On the Graphs page it is possible to configure a number of aggregations and graph configurations as follows:

Raw Data Sample

Allows configuration of the aggregation period and the aggregation function. The aggregation period can be:

  • None (Raw Data) - no aggregation is performed and the raw data is graphed. In this case the aggregation function is disabled.
  • Auto - one of the following aggregation periods are selected depending on the date range.
  • Hour - data is aggregated hourly using the selected aggregation function.
  • Day - data is aggregated daily using the selected aggregation function. The start of the day is determined by the time zone selected in the Account settings.
  • Week - data is aggregated by 7 day week using the selected aggregation function. Weeks run from Monday to through to Sunday.
  • Month - data is aggregated by Month. Due to the variable number of days in a month, comparisons between months may be inconsistent.

Due to daylight saving time, the length of a day, week or month can vary by an hour. The start of a day, week or month will align with midnight in the time zone set in the Account settings. If your browser is set to a different time zone, the graphs may not align to midnight.

To calculate aggregations, one of the following functions are used. Where there is no data sample within the aggregation period, values are extrapolated from the previous valid data point. This means that only changes in data need to be sent to the system rather than constantly polling (although that will work too). The aggregation function can be as follows:

  • Sample - the first valid data point within the aggregation period is used. Any other values between samples are completely ignored. This is useful for data that doesn't change much over the short term.
  • Change - calculates the difference between the first and last values in the aggregation period. This is useful for calculating the amount something has changed such as the amount of electricity used from two meter readings.
  • Sum - this adds up all the hourly averages during the period. This is ideal for calculating kWh from Watts.
  • Weighted Average - this calculates the average of the data values during the period taking into account the amount of time the data point was at that value. E.g. if a value was zero for 8 hours and 1 for 16 hours the weighted average for the day would be 0.667.
  • Minimum - the minimum data value found during the aggregation period. This is useful for plotting low temperatures
  • Maximum - the maximum data value found during the aggregation period. This is useful for plotting high temperatures
  • On Duration - the number of hours the variable was greater or equal to 1. This is useful for determining how long a device such as a light switch or heating controller was on during the day.
  • On Count - the number of times the variable changed from 0 to non-zero. This is useful for counting the number of times an infra-red sensor has been activated during a day, for example.
  • Sample Count - the number of real data samples received during the period. This may be zero if the data is extrapolated.

Column Chart

The graph types menu allows for selection of the following types:

  • None - hides the channel on the graph.
  • Icon Only - shows an icon of the channel - depending on the data type, the icon may represent the status of the channel.
  • Curved Line - joins the data points using a best fit curve.
  • Stepped Line - extrapolates each data point with a horizonal line to the next data point.
  • Scatter - plots a dot for each data point
  • Stepped Area - as stepped line but fills below the graph
  • Column - shows a column for each data point. Note that this may not make sense for raw data if it is not periodic.

The appearance of some of these options may be impacted by the graph options described later.

Show Range

The comparison menu will add an additional plot to the graph calculated from the 30th and 70th percentiles.

Percentiles are a highly useful technique for determining what the usual range of a variable is. For example to answer the question, am I using more or less electricity right now than I normally would at this time and this day of the week? The 30th percentile can be considered statistically as the low value for a data point - only 30% of the sampled data will fall below it. The 70th percentile can be considered statistically as the high value for a data point - only 30% of the sampled data will be above it. The mid value is the median where 50% of the sampled data is above the value and 50% is below. For the percentiles to give accurate results there must be enough data to sample, so if you have not been collecting data for long, the results may be variable.

  • None (Raw Data) - no comparison is added
  • Auto - one of the following comparisons are added depending on the date range.
  • Hour - Up to 13 weeks of hourly data are examined for each matching day of the week and hour of the day. E.g. for Sunday at 7pm - all the Sundays at 7pm for the past 13 weeks will be analysed.
  • Day - Up to 13 weeks of daily data are examined for each matching day of the week.
  • Week - Every week is examined for the entire data set.
  • Month - Every calendar month is examined for the entire data set.

The comparison menu itself has the following options:

  • No Comparison
  • Show Range - shows the higher and lower percentiles as a range over the data
  • Difference from High - subtracts the high (70th percentile) value from the data showing the difference between the data and the historically high values.
  • Difference from Mid - subtracts the mid (median or 50th percentile) value from the data showing the difference between the data and the historically average values.
  • Difference from Low - subtracts the low (30th percentile) value from the data showing the difference between the data and the historically average values.

Axis #1The axis menu is used to select the units that appear on the scale to the left of the graph. Selecting the same axis for two variables will cause them to share the same axis and scale on the graph. Axis #1 to Axis #3 are unitless labels that are available to allow up to 3 variables to be on a single graph, each scaling independently. A number of other common axes are also provided for convenience.

The units of an axis may be fixed for certain data types. For example, a channel monitoring energy usage may be automatically set to kWh. To change the units and axis of a fixed channel, the data type of the channel must be changed in the  channel config  menu.

The Options (gear) menu provides the following options which relate to the entire graph rather than the individual variables plotted:

  • Night time bands - adds daily shaded bands to the graph according to sunrise and sunset times. The bands are only shown when there is less than a month of data. The sunrise and sunset times are dependent on your latitude and longitude which can be configured in the Account settings.
  • Open at last data - when a graph is displayed on the dashboard, show the last available data recorded rather than the current date. For recent data, e.g. 1 day, this will show the last 24 hours of available data rather than the current day midnight to midnight.
  • Data markers - whether to display data markers (circles) for each data point on the spline and line chart types.
  • Legend - whether to display the legend at the bottom of the graph.
  • Stack charts - whether the charts on the same axis are stacked on each other or presented side by side. This is useful for adding units together such as the number of kilowatt hours a number of devices have used. Note that stacking charts on some axes, such as a temperature axis does not make much sense.
  • Wide lines - whether to use wider lines used to plot spline, line and area charts
  • Time axis scroll - loads additional data before and after the selected time range and allows scrolling on the x-axis. The amount of scrolling allowed is limited and may not cover the entire data range available. Use the calendar to select the required data range if it is not available via the scrolling mechanism.

The save icon allows the configured graph to be saved and added to the dashboard. To edit an existing graph, find it on the dashboard and edit it using the pencil icon. To remove a graph from the dashboard click the trash icon - this is only available if the graph was selected from the dashboard page.