When dealing with numbers and graphs, we hear two views. The analyst wonders why we even use graphs as the numbers are right in front of us. Meanwhile, the non-analyst throws his hands in the air and decides to rely on gut instinct. Neither of those options works. We’re surrounded by numbers, so much so that we can barely comprehend what’s happening. In order to spot trends and outliers, we need graphs.
Below are six simple graphs you can create using Excel-like products and in which situations you should use them:
Graph 1) Column
A column graph shows vertical bars that represent quantities. You would use this graph whenever you want to showcase values rather than the segments to which those values apply. Someone looking at the graph will see which columns reach highest or lowest and focus on those as the outliers.
Example: https:///img0.gmodules.com/ig/modules/column-chart.png
Graph 2) Bar
A bar chart is a column chart flipped horizontally. Because it focuses the viewer on the segments rather than the quantities, it is important to sort your segments by quantity when using a bar graph. Viewers will see a ranking of segments by the quantity you specify.
Graph 3) Line
A line graph connects all the points on an x-y axis, giving you a line that goes from left to right, jumping up or down depending on quantities. Normally you would use this when showing trends over time. This graph will show outliers over time rather than by segment.
Example: https:///edynblog.files.wordpress.com/2007/07/line-graph-days-on-market.jpg?w=600
Graph 4) Scatter
Scatter graphs plot discrete points of data on an x-y axis without connecting them. The viewer can identify trends without connecting the dots. Use scatter graphs where you have a lot of random-looking data, but you want to spot a possible trend.
Example: https:///www.optionetics.com/images/articles/3-19%20fig%201%20scatter%20plot.gif
Graph 5) Bubble
A bubble chart is similar to a scatter plot except that the size of the dot corresponds to a value like the number of records, amount of revenue. This is one of the few Excel graphs which allow the analyst to break out of two dimensions. For example, the x-axis can be revenue, the y-axis # of units and the size of the bubble could be the client company’s yearly revenue. Therefore you see how well revenue correlates with units and how both of those compare to the size of your customers’ revenue – so you’re looking at three dimensions instead of two.
Example: https:///flowingdata.com/wp-content/uploads/2010/11/5-edited-version1-575×385.png
Graph 6) Pie
Pie charts show the proportion of each segment in a population. The best use case is when a proportional threshold exists that you want to reach. For example, political coalitions must usually command more than 50% of seats in a parliament before they can pick a government. Therefore, pie charts are overwhelmingly used to show voter preferences in politics.
Example: https:///static.guim.co.uk/sys-images/Guardian/Pix/maps_and_graphs/2008/12/16/ICM_pie_460wide.gif
Multidimensional analysis
Visual analytics enables us to understand complex data at a glance. Because we can process so much information at once, simple bar and pie charts don’t use our brain’s full capacity. That means we can include a lot more data in multidimensional analyses, saving time, money and effort in understanding how our businesses work. For higher-level visual analytic software, check out Tableau Software or Tibco Spotfire
For a great historical example of a multi-dimensional graph, look at Charles Joseph Minard’s chart showing Napoleon’s march into (and out of) Russia from 1812-1813: https:///upload.wikimedia.org/wikipedia/commons/2/29/Minard.png
This graph shows three things: 1) the size of the Grand Armée as it marched through Russia, 2) the geographic position of the Grand Armée on its march and 3) the temperature soldiers had to face as they retreated from Moscow to Poland.
What is the best part of this graph? That it was designed to fit a specific objective. Minard wanted to show the cost of war to his fellow Frenchmen. Make sure you formulate the questions you want to answer before diving into dashboard creation and analysis.
