This tutorial will show you how to use the Trends panel inside your Graphext projects. The Trends panel offers a range of charts allowing you to plot the evolution of your data over time. To start using the Trends panel we need a dataset with a date variable and we need to build a project using it. The dataset we will use for this project is similar to the Fictional-Employees.csv file that we have used in other Getting Started tutorials. It contains information about 100 fictional employees and their characteristics. However, in the dataset Fictional Employees with Time, which we will use in this tutorial, there is an additional variable containing information about when the employee joined the company.
Once we’ve built the project, we will head straight over to the Trends panel and pick a type of chart to use. To begin, we'll use an Overview chart to examine how the number of employees joining the company has changed over time. Then, we’ll compare how the recruitment of High, Low and Medium performing employees changed over time before looking specifically at when the company hired employees with certain characteristics. Finally, we will use Share charts to get an impression of how values were proportionally distributed over time.
"The tipping point is that magic moment when an idea, trend or social behaviour crosses a threshold, tips, and spreads like wildfire."
- Malcolm Gladwell
To use trends charts, you must have a date or time variable in your data. There are 4 types of charts to explore inside the Trends panel, each offering a different perspective on your data over time. The y-axis represents a quantitative measure, which can be changed to reflect a count, average or sum of your values. The x-axis represents time. You can adjust the range of time presented in your charts as well and switching between the values presented on your axis. Additionally, you can annotate your trends charts to help provide context around peaks or troughs and mark significant events.
Let's quickly set up a project clustering our employees based on the similarity of the scores they received for our set of 10 characteristics. The project that we will build will include the same steps that were used in the Fictional Employees: Clustering project that we build in our other Setting Up tutorials.
First, make sure you have the Fictional-Employees-With-Time.csv file downloaded to your computer. Upload this to the workspace of your Personal team in Graphext and select it to bring up the project setup wizard on the right-hand side of the screen. For a more detailed tutorial on datasets, have a look at our Intro to Datasets tutorial.
Choose Employees as your analysis type and then Clustering. Then, inside of the Clusters and Network Creation menu, set the Performance level variable as your target and set the 10 characteristic variables as your factors.
Finally, inside the Network Visualisation menu, set Name as the variable to identify your nodes by and add Gender to the list of pinned variables. That's it. Continue to name your project something like Fictional Employees: Trends and execute it. Head back to the Projects panel of your Graphext workspace and hold on for a couple of minutes whilst Graphext builds your project.
For a more detailed tutorial on setting up a project, follow our tutorial Intro to Projects.
Starting from the Projects panel of your Graphext workspace, open the Fictional Employees: Trends project that you just built. Once it has opened, head straight over to the Trends panel of the project.
The window inside of your Trends panel will be prompting you to choose a type of chart. The dropdown menus have automatically been set so that your chart will visualize Node Count (a count of your nodes) on the y-axis and Date_joined (the date an employee joined the company) on the x-axis.
Additionally, there are 4 visualization cards representing the type of chart you want to use. Read more about these types of charts here. To begin, we want to look at an Overview chart visualizing how many employees joined the company for each year in the dataset. Choose Overview and select Go!
Your Trends panel will now be taken up with your very first chart; an Overview showing how many employees joined the company in each year of the dataset. Take a moment to look around the panel. There are many features here allowing you to filter and customize the appearance of data in your chart.
Taking up most of the room inside of your Trends panel is your Overview chart. This histogram makes it pretty easy to spot one thing in particular; that the company hired the most employees in 2012. Hovering over the 2012 bar will bring up the statistics behind this data point. It seems as though the company hired 20 people between January and April 2012.
The date range represented in the chart is currently set to quarterly. This is why the histogram represents data from January to April. Change this to a yearly range using the dropdown menu sitting above the chart to the left. Changing the date from a quarterly to a yearly range will result in your bars becoming very grouped together. Now, hover over the bar for 2012 and notice that the date range presented now reads 2012 - 2013.
Having your sidebar variable charts present in the Trends panel of your project means that you can filter the data presented in the charts. Currently, the chart visualizes data for all of the employees. Let's find out when the most top-performing employees were hired.
From the right sidebar, find the Performance Level variable and select the High value. The data in your chart will immediately change to reflect the active filter. Additionally, the scale represented on your y-axis will shift. Without the filter active your y-axis went all the way to 20 employees. With the filter representing High performer the y-axis only goes up to 8.
The gray histogram at the very bottom of your Trends panel allows you to select specific date ranges to visualize in your chart. First, make sure you clear all of your filters. Then, click on the grey bar representing data from 2013. Now you can drag the edges of the filter to cover a larger date range. Change the range inside your filter to represent data from 2013 - 2016. Then change the sample of your bars back to a quarterly representation.
You should see that hiring in the company grew in 2013, 2014 and 2015 but fell again in 2016. Clear your date filter and let's move on to look more closely at how values in our dataset evolved over time.
An Overview chart is a good way to get familiar with the distribution of your data over time but it isn't really giving us much insights about values in our data. Let's investigate how successful the company's recruitment was at different points in time by plotting the Performance Level variable in trends charts.
To compare the segments of Performance Level, we should use a type of chart allowing us to plot multiple values in the same chart. Using the dropdown menu at the top left of your Trends panel, switch the chart type to Compared Segments and change your sampling back to a yearly representation. That's better. Now your Trends panel shows a line graph with multiple values from the umap_cluster variable.
We want to plot values from Performance Level rather than umap_cluster. Find the variable dropdown underneath your main chart currently representing umap_cluster and change it to represent Performance Level. Once you've done this, your chart will show three lines, each representing a value from the Performance Level variable.
It seems as though the 2012 spike in the data is due to a relatively large number of High and Low performing employees being hired at that time. The green line shows that recruitment of Medium performing employees dipped during the same period but rose between 2014 and 2016.
Currently, our chart represents the number of employees that were hired. Let's dive a little deeper into discovering what kind of employees were hired at different times. Instead of measuring a count, we will reconfigure the chart to represent Intelligence on the y-axis. The aim here is to discover if a trend exists in the company's hiring of High performing, intelligent employees.
Click the x icon next to the Low and Medium values so that the blue line representing High performing employees is the only one in the chart. From the top of the panel, above the trends chart, select the dropdown menu currently representing Node count. Change the value so that your chart represents Intelligence on the y-axis.
After you've done this, notice that the y-axis scale has changed to represent an Average rather than a Count. This is because the axis represents the average Intelligence score of all employees that were hired during that year.
This is quite a revealing chart. It seems that there is quite an evolution in the number of High performing, intelligent employees that were hired by the company. During the first years of the dataset, the average Intelligence score of High performing employees is consistently below 5. After 2014, this rises dramatically.
Annotating trends charts with text describing significant events can help to add context to your data. Since the rise in recruitment of intelligent, High performing employees takes place at a specific moment in the dataset, it makes sense to mark this on our chart. Click the pin icon representing the ability to add an annotation from the top right of your Trends panel. Now, move your cursor over the point in the chart where the rise takes place and click again.
After you've identified the point to add an annotation, a vertical line will be drawn onto your chart marking that point. Click on the New annotation text and change the text so that your annotation reads; Company hires more Intelligent, High Performers.
Now change the chart back so that it represents a count of all 3 of your values from the Performance Level variable.
Share charts offer something a little different. In Share charts, values belonging to specific variables are divided into percentage shares, showing how each value contributes to the features of a variable over a date range.
Let's use a share chart to visualize how the proportion of High, Medium and Low performers recruited by the company changed over time. To switch your chart type to a Share chart, click on the chart dropdown from the top left of your Trends panel and choose Share from the menu list.
Once you've changed your type of chart, take a look at the distribution of values of the y-axis. These now represent a percentage value. Share charts always show how your values contribute to 100% of the variable that is being visualized. The spikes that we observed using the Compared Segments chart is also visible here.
An interesting aspect of this visualisation is the appearance of correlation between the recruitment of High and Low performing employees. These values follow quite similar trend lines, when visualized in a Share chart, whereas Medium performing employees don't.
We know that data isn't always clean and simple.
Have a look through these topics if you can't see what you are looking for.