Difference between revisions of "IS480 Team wiki: 2017T1 Noir Knights Final Wiki"
(updated link 4) |
|||
(4 intermediate revisions by the same user not shown) | |||
Line 284: | Line 284: | ||
* Objectives:<br> | * Objectives:<br> | ||
To test which algorithm provides the best improvement after 100 assessment questions are done. | To test which algorithm provides the best improvement after 100 assessment questions are done. | ||
− | [[IS480 Team wiki: 2017T1 Noir Knights AB Testing|View our | + | [[IS480 Team wiki: 2017T1 Noir Knights AB Testing|View our AB testing results here!]] |
<h4>User Acceptance Testing 1</h4> | <h4>User Acceptance Testing 1</h4> | ||
Line 329: | Line 329: | ||
To improve web application based on UAT results | To improve web application based on UAT results | ||
[[IS480 Team wiki: 2017T1 Noir Knights UAT 4|View our UAT4 results here!]] | [[IS480 Team wiki: 2017T1 Noir Knights UAT 4|View our UAT4 results here!]] | ||
+ | |||
+ | <h4>A/B Testing</h4> | ||
+ | *Venue: St Hilda Primary School (Main) | ||
+ | *Date: 17th August 2017 | ||
+ | *Participants: 18 | ||
+ | |||
+ | *Objectives:<br> | ||
+ | To test which algorithm provides the best improvement after 100 assessment questions are done. | ||
+ | |||
+ | The following 3 tasks are performed by the testers 1 week before the actual testing date: | ||
+ | Task 1: Registration | ||
+ | Task 2: Attempt Diagnosis Test | ||
+ | Task 3: Attempt Assessment questions (Algorithm are allocated without their knowledge) | ||
+ | |||
+ | The following Task is performed by the testers upon the actual testing date: | ||
+ | Task 4:Attempt exam | ||
+ | |||
+ | Click [[IS480 Team wiki: 2017T1 Noir Knights AB Testing|here]] for the A/B Testing results | ||
==<div style="background: #000000; padding: 15px; font-weight: bold; line-height: 0.3em; text-indent: 15px; text-transform:lowercase;letter-spacing:-0.08em;font-size:20px"><font color=#FFFFF>Reflections</font></div>== | ==<div style="background: #000000; padding: 15px; font-weight: bold; line-height: 0.3em; text-indent: 15px; text-transform:lowercase;letter-spacing:-0.08em;font-size:20px"><font color=#FFFFF>Reflections</font></div>== | ||
[[File:Nk_finalsr.png|1000px|center|link=]] | [[File:Nk_finalsr.png|1000px|center|link=]] |
Latest revision as of 00:12, 18 November 2017
Main Wiki | Midterms Wiki | Final Wiki |
Contents
|
Project Management
Project Scope
Acceptance | Midterms |
---|---|
Final Completed Scope
Major Scope changes
- Movement of payment module and discussion board as well as flagging to good to have functions as clients wants to add more specification and for us to focus more on the exam and report module
Project Schedule
View detailed Project Schedule here!
Planned | Actual |
---|---|
Schedule Highlights
- Removing of UAT1 due to it being an internal UAT as advised by Prof Ben during acceptance
- Renaming of UAT1, UAT2, UAT3, UAT4
- Rescheduling of UAT1 and UAT2 due to client's availability, as we are utilising Basecamp Learning Centre students.
- Including A/B testing to check on algorithm effectiveness
Metrics
Task Metrics
Task Metrics Highlights
Iteration | Task Metric Score | Action | Status |
---|---|---|---|
4 | 90% | Estimates are generally accurate and on track. Proceed as per normal.
Delayed slightly due to uploading of pictures to DB, team was unfamiliar with cloudinary and hence took a longer time to upload the pictures. Follow up action: Task was pushed to the next iteration. |
Completed |
5 | 78% | Need for greater efficiency. Re-estimate time needed to perform tasks and consider delaying tasks if necessary.
Delayed slightly due to deployment of script onto AWS. Pushing back deployment and testing and debugging of application Follow up action: Task was pushed to the next iteration. |
Completed |
7 | 90% | Estimates are generally accurate and on track. Proceed as per normal.
Delayed slightly due to bug of assessment page not showing the correct question that the machine learning algorithm sends. Follow up action: Task was pushed to the next iteration. |
Completed |
9 | 86% | Estimates are generally accurate and on track. Proceed as per normal.
Realised the need to use spring security for email verification. Emailing of report pushed to next iteration. More research to be done for implementing spring security. Follow up action: Task was pushed to the next iteration. |
Completed |
11 | 89% | Estimates are generally accurate and on track. Proceed as per normal.
After meeting with Prof Hoi, team decided to develop a separate platform for A/B testing to ensure that testing user's data is singled out. Follow up action: Task of history page pushed to next iteration. |
Completed |
Bug Metrics
Bug Score
Bug Distribution based on Severity
Iteration | Bug Score | Summary of bugs | Action Taken |
---|---|---|---|
5 | 12 2 High 2 Low |
Most of the bugs in this iteration were due to the Deployment of Python script. | Stop current development and resolve the bug immediately. Project Manager reschedules the project. |
9 | 14 2 High 4 Low |
MAB not generating questions. Radar chart not showing. |
Stop current development and resolve the bug immediately. Project Manager reschedules the project. |
Project Risks
Activated Risks
View our risk assessment here! The below table only highlights the two top risks that have been mitigated.
No | Risk Description | Impact | Likelihood | Impact | Mitigation Plan | Status |
---|---|---|---|---|---|---|
1 | Team has different priorities during project submissions period | Project schedule will be affected | Medium | Medium | Team to sit down and come out with a plan and assigned day to code, integrate and debug. | Mitigated |
2 | Unable to get Primary 6 students for testing due to their curriculum and proximity to PSLE. | Unable to get genuine feedback from students and hence affect the usability of the system | High | High | PM to reschedule project to push forward or arrange tasks to fit student's availability for testing. | Mitigated |
The above two risks has been mitigated.
- Risk 1: The team has worked up a schedule according to the different module deadlines and allocated days to code, integrate and debug.
- Risk 2: PM rescheduled the project to complete functions that are more important for testing, such as the assessment and exam modules to suit the students' curriculum.
Future Risks
No | Risk Description | Impact | Likelihood | Impact | Mitigation Plan | Status |
---|---|---|---|---|---|---|
1 | Unexpected issues when users use the system | Inconvenience will be caused for the users | Low | Low | Upon handover, the team will be giving the client a 1 month warranty period whereby all bugs found within this 1 month of usage will be solved by the team | Mitigated |
The above risk have not occurred. However, our team anticipated the possibility of it and managed to mitigate these risk.
Technical Complexity
Scoring Algorithm
- Overwrite differs from traditional marking systems whereby all students are given fixed timing on a certain set of questions
- The scoring algorithm aims to accurately measure student’s competency for every question and documents student’s data trail precisely
- As seen from the diagram, the app first gets factors such as time taken, level of difficulty after a student does a question.
- The app queries the database for information such as average time taken of question by all students and number of questions answered correctly consecutively.
- Using these 4 factors and the algorithm, a score for that question will be calculated. We will also update an average time taken of the question by all students and update the number of questions answered correctly consecutively.
Reinforcement Learning
- Using a purely arbitrary scoring algorithm mentioned above is not a comprehensive solution in determining a student’s competency in a certain subject with many topics.
- A holistic and dynamic testing environment is required to better analyse student’s full capabilities and imitate an examination condition
- Utilising Multi-Armed Bandit Algorithm to predict student’s weak topics within the shortest number of trials possible
- Every question answered by student will run the Upper Confidence Bound 1 algorithm.
- Algorithm would evaluate student’s performance and rank all topics in terms of their growth potential and likely hood of generating the most ‘rewards’ (weakest topic)
- Machine would then decide between ‘exploration’ vs ‘exploitation’ in determining whether to generate a question of a different topic or proceed with testing the same topic
- This interactive element between student and machine for each question allows for reinforcement learning to occur as machine would be able to better predict student’s strength and weakness more accurately after each round
Deployment
Deployed Link: http://54.213.80.240/overwritemaven/
Testing
We have conducted One A/B testing and Four User Acceptance testing.
A/B Testing 1
- Venue: St. Hilda's Primary School
- Date: 29th October 2017 - 5th November 2017
- Participants: Primary 6 students
- Number of Participants: 18
- Objectives:
To test which algorithm provides the best improvement after 100 assessment questions are done. View our AB testing results here!
User Acceptance Testing 1
- Venue: Basecamp Learning Centre
- Date: 17th August 2017
- Participants: Secondary 4 students
- Number of Participants: 3
- Objectives:
To gather feedback regarding user interface of developed functions from prospective users
To detect usability issues based on user behavior
To improve web application based on UAT results
View our UAT1 results here!
User Acceptance Testing 2
- Venue: Basecamp Learning Centre
- Date: 12th September 2017
- Participants: Secondary 1 students
- Number of Participants: 2
- Objectives:
To gather feedback regarding user interface of developed functions from prospective users
To detect usability issues based on user behavior
To improve web application based on UAT results
View our UAT2 results here!
User Acceptance Testing 3
- Venue: SIS CR3-1
- Date: 29th September 2017
- Participants: SIS students
- Number of Participants: 14
- Objectives:
To gather feedback regarding user interface of developed functions from prospective users
To detect usability issues based on user behavior
To improve web application based on UAT results
View our UAT3 results here!
User Acceptance Testing 4
- Venue: Different homes/ schools
- Date: 10th November 2017 - 14th November 2017
- Participants: Primary 6 students
- Number of Participants: 14
- Objectives:
To gather feedback regarding user interface of developed functions from prospective users
To detect usability issues based on user behavior
To improve web application based on UAT results
View our UAT4 results here!
A/B Testing
- Venue: St Hilda Primary School (Main)
- Date: 17th August 2017
- Participants: 18
- Objectives:
To test which algorithm provides the best improvement after 100 assessment questions are done.
The following 3 tasks are performed by the testers 1 week before the actual testing date: Task 1: Registration Task 2: Attempt Diagnosis Test Task 3: Attempt Assessment questions (Algorithm are allocated without their knowledge)
The following Task is performed by the testers upon the actual testing date: Task 4:Attempt exam
Click here for the A/B Testing results