1. Types of reports
Welcome to the third chapter! So far, we learned the importance of storytelling to impact the decision-making process,
the steps on the data storytelling road
2. Presentation strategy
and the elements to consider when choosing a suitable format to deliver our results.
3. Chapter 3
In this chapter, we will learn how to structure a written report.
We'll learn when to use several types of reports,
how to achieve reproducibility in data science,
and methods to write precise and clear reports.
4. Written reports
Imagine that at communicatb, we need to write a report
explaining our project focused on
sentiment analysis of product reviews.
5. Written reports
Our report should communicate our findings
regarding the negative ratings due to delayed shippings,
and the prediction accuracy our model displayed. We need to present
our work according to the industry standards
6. Written reports
so our results are validated, and our recommendations drive the expected change.
7. Types of reports
There are different types of reports.
Informational reports
provide factual information,
are usually short,
don't have a strict structure,
and their purpose is to inform about facts without adding any analysis.
Analytical reports
provide analysis, and demonstrates relationships or recommendations.
They can vary in size,
but have a strict structure,
and their purpose is to drive decision based on data.
8. Final report
Final reports
include detailed data analyses, findings and results as well as visuals.
They are usually long,
because they are intended for audiences that need technical details.
9. Summary report
Summary reports
include key findings and recommendations, as well as visuals.
They are usually short, less than five pages long.
Because they are a summary of a final report, they can include links to the main document.
They are intended for decision-makers that do not need technical details.
10. Report structure
As we mentioned informational reports do not follow a strict structure. But there is a straightforward structure we should follow for analytical, final and summary reports.
The first section is the introduction.
Here, we summarize the purpose of the report.
In our example, the purpose of the report is to show the analysis on the product reviews,
and the results of the rating predictions.
After that, we need to include contextual information about why we performed the analysis shown.
What motivated our example report is an increase in negative reviews.
Finally, we should summarize our analysis questions.
We ask which factors are driving the bad user experience.
11. Report structure
Then, we create the sections for the body of the report.
In the data section, we write a description of the most relevant data used. We can include tables.
In the Method section, we describe the methods used to gather and analyze the data, and build the model.
In our case, we explain we used natural language processing and a random forest model.
In the Analysis section, we include the selected data for the analysis and model using visuals,
for example, a graph with the most common words.
In the results section, we describe and explain the results of our analysis. We could also include visuals to evaluate them.
For example, we state that 30% of the negative ratings were associated with the terms "delayed" and "shipping".
12. Report structure
Finally, we create
a conclusion section. Here,
we restate the analysis questions,
as well as summarize the most important results from the analysis.
This part is the appropriate location to add our recommendations for the next steps.
13. Report structure
The structure we just saw is well-suited for data reports in journals.
In a business context, your audience is different so you need to adapt. An efficient method is the 1-3-25 approach:
1 page of abstract,
3 pages max of executive summary
and 25 pages max of details.
The executive summary is enough for people to understand your conclusions and decide whether the report is worth reading in its entirety.
14. Audience
Again, we should keep our audience in mind. Each stakeholder is interested in different parts of our report. So we should tailor it accordingly.
For example, people with little time but interested in the topic will read the introduction and conclusion and then scan the body for particular points.
15. Audience
The executive team will quickly read the introduction and the conclusion to find clues related to their main interest: our recommendations.
16. Audience
Finally, a technical stakeholder is mainly interested in the body of the report. They want to understand and validate our technical methods and analysis.
17. Let's practice!
Now, it's your time to write a report!