Time Series Analyses

Background

Some labs capture results over a period of time for tests on the same Sample,  studying and interpreting a sequence of data points recorded or collected at consistent time intervals e.g. ambient lab temperature, weather records, patient health evolution or physical strength tests in engineering

Consider an example where the percentage mass change of 3 similar drying concrete bars are regularly measured, tabled and eventually graphed including of their average

Set up the Time Series Importer

After it was set up, and the Analysis Service, see overleaf

Using Time Series Analyses...

Graphing yourn results

Colours

These are from the Tableau 10 palette, designed for readability and data visualization. They are balanced so each color is distinct against white and against each other

Teal #17becf

Olive #bcbd22

Gray #7f7f7f

Pink #e377c2

Brown #8c564b

Purple #9467bd

Red #d62728

Green #2ca02c

Orange #ff7f0e

Blue #1f77b4

Setting up a Time Series Analysis Service

For the given example to Analysis Services, and Crete a new one. After completing all its standard attributes, open its Results Options tab

Select Time Series as a result Type

Graph Smoothing and Colours

You can also select the type of interpolation you want

Basis  smooth, the line does not necessarily pass through the given data points

Cardinal smooth, the curve always passes through the given data points

Linear spiky, connects each pair of points with a straight line

Data Table Definition

Proceed to lay out the graph by defining its columns and an Index to which the graph will be plotted to - the x-axis, drying cycles in this example

NB  There has to be an Index row, select it on the Column Type lookup which will be the X axis units in the spreadsheet to be imported for the Analysis

Follow up with the data columns, these are the lines that will be graphed - give each the Data column type and appropriate title

If the value is a whole number, select Number for the Data Type, Float if it is a rational number that includes a fraction

If it is an Average, select that as Column type - it will be used to differentiate it in colour from the other data lines

You may also add Error Bars to Average curves, and table some results and Hide them without plotting them

Assign custom colours to lines in the right hand column. If no colour is given, the line won't be drawn and just it's data points plotted