Long Time Series

Trend analysis is by far the most used type of analysis and presentation style in a dashboard or report. Time series trend data hasTrend_Sizes_1 different levels of granularity and the most used time slices are i.e. Day, Week, Month, Quarter and Year. There are sceanrios where user looks at the data every minute, for example stock price data, but it is used only in specialized occasions.

A typical time series representation will show couple of years of data summarized as per the desired granularity (e.g. 5 years, 8 quarters, 13 months, 22 weeks or 30 days etc). Majority of requirements will be in the range of 4 to 30 data points to be plotted in a chart. In circumstances where a large set of data points have to be represented in time series fashion, what is the best approach for the layout of a chart. For example, user has requested to show the number of widgets manufactured every day by a factory in the last one year. Here, the number of data points are 365 and the chart will look cluttered if we blindly follow prior design principles.

In abnormally large data sets like the scenario mentioned above, the user request should be treated with a different approach. A request like this can be interpretted in two ways where the user has the following thought trail in his/her mind

  1. I want to look at the data points where there are upward and downward spikes explicltly and further probe from there on.
  2. I want to look at the general trend and see the future in terms of data movement. 

Based on the users thought process, the chart size has to be proposed in the wireframe. Adjusting the Height vs Width ratio can highlight and provide a clear picture on end-user expectations. For the first scenario, if the user wants to analyze peaks and valleys, the ideal size should have a 2:1 ratio of height against width and for second scenario of general flow analysis, the ideal size would be 1:2. Although the final size depends on quantum of real-estate available to display data, this is a general guide to keep sizes consistent with type of analysis to be performed.


A random plot above of the same dataset with two different layouts as mentioned previously. It is evident that based on size of the layout user attention can be taken to the desired area. The above sizes are indicative only and as to goes in any project, final stages involve tweaking the layout required in order to meet the expectations.