Testing

  1. Home
  2. Testing

Testing Strategy

Due to the specific user group and clients requirements, a significant portion of our code, such as that used for plotting graphs/tables and downloading data from DesInventar cannot be tested with unit tests.

Therefore we relied primarily on manual testing by hand tracing, comparing estimation of the expected behaviours or manually calculated result with program output. For significant unexpected results we investigate both our code and the data accrue the error.

Besides, we weekly report our results to our clients and they are validated IFRC's data scientists.

User Acceptance Testing

To ensure CIDAS is understandable and works as expected of our clients, we invited 4 people including but not only include our project partners to test our tools, as a simulation situation of our persona.


To ensure anonymity and confidentiality, testers are masked with portraits and names listed in persona.



Tester ID Tester Name Tester Role
1 Green Somebody Treasury Department Officer
2 Graham Ginnetti Chief Executive Officer
3 Justin Roberts IFRC Software Engineer
4 Glass Blue Data analyst

We designed 4 typical tasks users may want to achieve with CIDAS. Testers would give custom comments after filling the Likert Scale form.


Test Case ID Description
1 Update database. Once our data source including DesInventar updates their data, we may want to update our analysis as well. Testers are asked to use the data-downloader tool to download the latest data.
2 Reprocess data. Testers are asked to process downloaded data with the data-processor tool.
3 Testers are asked to generate the flood return period-fatalities graphs of Albania and Pakistan with the data-visualiser tool.
4 Play around. Without specific detailed purpose, check the return period graph or table of any country with any slicing rule interested.
Acceptance Requirement Total Disagree Disagree Neutral Agree Total Agree Comments
CIDAS produces the result without error 0 0 0 1 3 +The code is robust and the workflow is repeatable
-Data processor need some adjustment to run the example
The guidance and process is intuitive and straightforward 0 0 1 1 2 +Every step is well documented with detailed instructions
-When combining everything it becomes a little too complicated
CIDAS response to every user command 0 1 0 0 3 +Most of the tools provide clear progress notice
-Data processor tool only responds when an error occurs
Fast response 0 0 1 0 3 +Plotting graphs is instant
-Data processor tool response relatively slow
Clear and well-organized output 0 0 0 0 4 N/A
Accessible on different environments 0 0 0 1 3 +Can access from Windows, Mac OS or GitHub codespace
-Data processor needs adjustment to run on codespace
Can do multiple jobs sequentially 0 0 0 0 4 +Can plot graphs with flexible amount of countries
-Have no easy access to losses of all events for all countries
From the User Acceptance Test we can conclude that users are generally satisfied with the output of our system, but can be further improved on the workflow by simplified some processes and adding notice of progress for some tools, and easier way to access all results.

Considering the frequency of potential usage, this system would not need to be deployed and run very frequently, therefore CIDAS' performance is generally acceptable and we are glad to confirm we have received very positive feedback from our clients.