Creating a Line Plot

Overview

The line plot facilitates the visualisation of data across related properties (commonly all properties will come from the same analysis group). A key difference from the similar parallel coordinate plot is that all properties shown on the plot must share a common indicator, unit group and ratio group, as they will all be plotted on one, shared axis.

Version: 1.17+ (Apr 2019)

Usage: Graph --> New Graph --> New line plot...

How to use in practice

Line Plot creation

The line plot creation dialogue requires the selection of a group of numerical (geochemical) properties. A minimum of two, and a maximum of 60 properties is required for creation. We recommend that 20 or so properties be considered a sensible limit for visualisation, but permit up to 60.

A key requirement is these properties all share the same indicator, unit group and ratio group. This ensures the results can all be plotted on the same axis. The create dialogue includes validation to ensure only permissible line plots will be created and will let you know which properties do not fit. The indicator, unit group and ratio group is defined based on the property at the top of the list of Selected properties.

Properties are selected using the standard list based Property Selector present throughout the software, and can be searched for in the Available Properties list, and drag and dropped into the Included Properties list. The order of the properties shown on the graph is determined by their order in the Included Properties list. A count of the number of included properties is provided. By default properties are included in their default indicator ([un] where applicable), units, and ratio.

Interaction with the Line plot

The line plot has similar interaction to other graphs, but with some specific custom behaviour outlined below.

Once created the shared line plot axis scale and indicators, units and ratios can be managed through the Axes tab in Graph Manager. Remember this shared axis applies to all properties on the plot. We can invert this shared axis, or when the range is positive, apply a log scaling. The axis label is by default defined to say "Value[indicator] (unit)" but can be modified. The indicator will only be shown if it is relevant.

The individual property axes can have their naming labels modified to create a plot to the users requirements using the Axes box below. Note the actual label text is shown in the Label box as each label is selected in the list, which shows the property and analysis short names.

Version: 2.1 We altered the behaviour of the shared axis label on line plot creation so that they now show only the properties on the axis labels by default helping the graph use space more effectively. A button allows the user to append the analysis names if desired.

Adding/removing properties from the graph is accessed through the right click menu from the graph. Use is similar to the generic bulk selection of properties dialogue. It is only possible to add numerical properties which have consistent indicator, ratio and unit groups to a line plot (due to the shared y-axis). The first property in the list is used to define the permissible indicators, units and ratios as in the create stage. Users will be informed when inconsistent properties are added. When at least one existing property is retained the shared axis extent is retained, and it may be necessary to zoom to data to see the values for that axis.

Bulk update of analysis groups, indicators and units is possible on the line graph in two ways. First, the user can change the indicators and units on the Axes tab, which will update all properties on the graph as this is for the shared y-axis. Secondly the user can select the right-click option to Bulk update properties option on the graph. This brings up a slightly modified bulk update dialogue. Here you can only select target analysis, indicator or unit, as by construction the graph will already share the same indicator and units. This allows you to quickly change e.g. from looking at Sat-GC to WO-GC, or from [h] to [a] indicators, or from ppm to ppb across all properties. In general it is probably simpler to use the Axes tab version for indicator and unit changes.

The Exploration tab allows you to select which interaction mode to have with the on-screen graph. By default you will be in Zoom interaction mode. This allows you to Zoom to data, which will zoom the shared axis to the data range of all the data to be shown on the plot. You can also Save and Restore the zoom.

It is possible to select samples on the line plot using Brushing - note that brushing selects the sample, so selecting one point will select the whole 'line' for that sample. As with other brushing options it is possible to clear the selection, open the Visual Query tool to view the sample information or add the selected samples to a static sample set.

The Layout tab allows limited interaction on the line plot. You can edit the graph title, turn its display on/off, select to show/hide the property axes (the grey lines that guide the eye at each property axis), show/hide the legend and select to show/hide the data points and lines.

An additional option on the line plot is the Data tab. This allows the user to select the normalisation applied to the line plot data.

By default no normalisation is applied to a line plot. The user can select from three different options for data normalisation. Each acts across the sample, that is the normalisation is applied to the properties selected to be shown on the line plot, on a per sample basis. The normalisation options are:

Range normalise: this scales each sample so that the maximum is 1, and the minimum is 0. Missing data is simply ignored. If only one value is present this is set to 0.5.
Area normalise: this scales each sample so that the sum of the absolute values scales to 1. For positive quantities, such as GCxx data this is probably the most natural normalisation to use. Missing data is treated by reducing the sum from 1 to n/N where n is the number of properties with data, and N the total number of properties on the plot.
Statistical normalise: this scales each sample such that the sample has zero mean and unit variance (or standard deviation). If there are missing values these are treated as usual in the calculation of the mean and variance (we scale by the population calculations, i.e. a factor n). If there is one value we set this to 0.

Normalisation can be very effective in bringing data to a common scale, but should be used with care. It is a linear transformation of the data, but it operates on a per sample basis. The interaction is immediate so the user can easily gauge the effect of changing the normalisation options on their perception of the data. It is most usefully applied to GC and GCMS data where this is supplied as heights or areas, to remove the effect of the arbitrary measurement scale and lack of knowledge of the quantity of substance injected into the instrument.

Graph interpretative layers are not currently supported on line plots.

If you wish to share your configured line plots with colleagues, you can simply right click on the line plot in the artefact tree and select to Export As Template. This will produce a template (without data), but with all the settings you have defined for the line plot, that can readily be applied to other projects.