Communicating findings

Static visualisation

After conducting analysis and plotting various plots for comparisons, the next step would be to effectively communicate your results.

The next Jupyter Notebook picks up with generating the final visualisations used to communicate our findings. In the notebook follow the code snippets and observe your static visualisations coming live.

Download: Static-visualisation-data.ipynb

Interactive visualisation

Finally, there’s also brazil_fire_bokeh.py for you to download, which is a Bokeh dashboard showing a time series plot of the historic fires per month.

Download: brazil_fire_bokeh.py

If you notice the code snippet, we use a DateRangeSilder to allow users to select the range of data they would like to see – this is like a slider but is specifically built to show its values as dates. We also add interactivity using the value_throttled attribute for on_change. This means the new date values are only sent when the user finishes dragging the slider, otherwise we can be overwhelmed with requests.

The dashboard can be run in the usual manner, using the command:

bokeh serve --show brazil_fire_bokeh.py

Attributes of actionable insights

You might not always get actionable insights, but it is good to know how you can decide whether your insights are attainable or not.

Alignment

Link the insights to the organisation’s goals and strategic objectives. Insights that are based on KPIs and other key metrics create a sense of urgency. It’s easier to interpret and convert strategically aligned insights into tactical responses because such insights are directly related to your business goals.

Context

Provide a benchmark or comparison to give your data useful context. It’s difficult to move forward if you lack the context to explain why the insight is unique.

Relevance

Deliver the insight to the right person, at the right time, in the right context. Communicate insights to relevant decision-makers so they get the attention they need. Make sure the messages are timely so insights are not out of date when stakeholders act on them. Capture insights in an analytics tool that managers can access and use so the message reaches the intended audience.

Specificity

If an insight is specific, it’s more likely to be acted on. Insights based on KPIs and other high-level metrics can sometimes highlight interesting anomalies but lack enough detail to foster immediate action. If there’s not enough information or explanation about an insight, it’s not yet actionable. You need to probe deeper into the insight before acting on it.

Novelty

A novel insight is more interesting than a familiar insight. A novel insight encourages leaders to verify and test a finding that could shed light on new opportunities or challenges.

Clarity

Communicate insights clearly to foster adoption and success. If users understand why an insight is important and how it can help them, they are likely to act.

You now have a strong foundation for analysing data and presenting your findings in visually pleasing and easily understandable ways. You should be able to take data from a number of sources and formats, clean it, then present it in a way that tells a story to your audience.

What do you think?

In the Brazilian fire scenario, we don’t really continue into the final stage of iterating. However, there are some paths we could have taken, such as looking at climate data for historical months. To see if that can cause more fires, we might need more domain knowledge so that we can answer further questions. For example, are there bigger fires where they can cross state boundaries, leading to a snowball effect for the number of fires?

How do you think we can approach visualising such analysis?

Share your thoughts with your fellow learners in the comments below.