1. Case study: report on credit risk
Good job! Let's go through a small case study now.
2. Credit risk
We're going to decide step by step how to write a report on a credit risk case. So let's present the case.
Credit risk quantifies the probability that a customer fails to meet contractual obligations, such as credit card debts, and other loans.
The finance team of a fictional Bank, Loanme, is interested in understanding and predicting which customer taking on a loan is more likely to default.
They send us data available on loan details and borrowers, including age, income, loan amount,
that we use to perform data exploration,
and to train and evaluate a predictive model. Now it's time to show our results.
3. Audience
We are going to communicate our findings to non-technical stakeholders,
mostly to decision-makers.
4. Story
So first, let's find our story, and especially our narrative.
There was an increase in the percentage of defaulting customers over the last 5 years. So the bank became interested in predicting which customers had a high probability of default.
We analyzed the data and saw that people with more unemployment periods tend to default more.
Also, our analysis showed that younger people with less income tend to default more.
After training our model, we saw that it is possible to predict which people are more likely to default with an accuracy of 95%.
To confirm these results, the next step should be to run a trial on a controlled population.
5. Tech or non-tech
We have our story. Now we need to translate technical results for non-technical stakeholders.
6. The right data
Let's define an
audience persona.
We report to the finance department director,
who needs to decide on implementing an automated loan rejection system using our project findings.
So we report the relationship between age or income and loan default
and forecast percentage of customer defaulting over the next months.
7. Statistics
Because we need to summarize numerical data into an aggregate, we show the median age and income for default versus non default customers.
Also, we need to show how the number of customer defaulting changes over time, so we show the percentage of change.
8. Visuals
Accordingly, we include a boxplot showing age or income vs. default condition,
9. Visuals
and a lineplot with the percentage of change in defaulting customers over the next months.
10. Correct format
We now have all our pieces together. Let's recap.
We are reporting to the Financial Department director,
because she has an important decision ahead.
She is interested in our key findings and recommendations,
and we should communicate the findings through email before our meeting.
11. Report
All of these elements, lead us to select a written report to deliver our results.
But should we select a summary or a final report?
12. Report
It's a non-technical stakeholder, so a summary report should be better.
Should it be an informational or an analytical report?
13. Report
We are presenting not only facts but also our analysis. So the most suitable is the analytical report.
14. Summary report structure
So how are structuring the summary report? Let's see.
In the introduction,
we first summarize the purpose of our report,
we add contextual information about the reason of our project,
and state our analysis question.
Then, in the body, we are going to
describe the data,
and include only key findings in the result section.
Lastly, we restate the questions linking it to the central insight
and we add our recommendations.
15. Let's practice!
Now, it's your turn to organize summary reports for stakeholders!