diff --git a/documentation/T1_2024/Group_Tasks_and_Handover_Documents/DataBytesReport.md b/documentation/T1_2024/Group_Tasks_and_Handover_Documents/DataBytesReport.md index 6c5b783f..90360160 100644 --- a/documentation/T1_2024/Group_Tasks_and_Handover_Documents/DataBytesReport.md +++ b/documentation/T1_2024/Group_Tasks_and_Handover_Documents/DataBytesReport.md @@ -422,5 +422,6 @@ height="4.840277777777778in"} | Aishwarya Mahajan |
  • Proposed and established a comprehensive data quality platform by sharing a data quality report.
  • Created validation scripts for data dictionary ensuring data integrity and accuracy for customer, account, transaction and loan tables.
  • Implemented checks for not-null constraints, unique constraints, range validations, allowed values, and regex pattern matching.
  • Introduced a new Data Quality folder in the GitHub repository.
  • Formed and led a dedicated Data Quality team focussing the validation of transactional data in database.
  • Developed Python scripts for visualizing monthly spending data for users.
  • Created detailed bar charts with color gradients to indicate spending intensity.
    • | | (Mariska) Heera Mohanadas |
  • Reviewed logging function, created report / review on it.
  • Enhanced logging function, adding log rotation and reduced noise
  • Created internal denial list to reject common/weak strings in potential passwords
  • Conducted risk assessments on two third-party services, GroqCloud API and Pinecone
  • Attended leadership meetings to discuss handover documents, and discuss project future
  • Created script for Dolfin (old) showcase video, recorded my part and then added Junkai's part into it, and then uploaded showcase video to Panopto
  • Reviewed the work of other students (code and documentation) to ensure they satisfied corresponding requirements of their Trello task, primarily cyber security students though
  • Answered student queries
  • Regularly posted updates and notified Dolfin project members on the timelines of the unit
  • Answered student queries, troubleshooting
  • Conducted code reviews on PRs, ensuring they do not adversely affect or compromise the security of the website or its credentials
  • Hosted a meeting to discuss handover documents and delegation with junior leadership students
    • | | Denica Hope |
  • Finalised the creation of the data generation python scripts for transaction data, home loan data, credit card data including compiling a reference list of hundred Australian merchant names (real businesses) that I also categorised with our new classfication categories
  • Created a python script to replicate item 3 of the Melbourne Institute-CBA Observed Finance Wellbeing Scale and tested against multiple datasets to validate output was correct
  • Created gauge chart using python to visualise the years remaining on a home loan (displayed loan total years, years passed and years remaining)
  • Created a sankey chart using python to visualise the spread of total spend across the mid level transaction classification heirachy
  • Held weekly meetings for DS/AI stream, created group posts of key dates for end of trimester, created new trello cards for visualisations and financial wellbeing tasks, assisted students with clarifying tasks, providing feedback on submissions
    • -| Gimsara Kenula Elgiriyage |
    1. Met with the Product Owner and mentor weekly to update and discuss the product and tasks.
    2. Created new Trello cards for Front End stream.
    3. Assigned tasks to FE team members.
    4. Hosted meetings with the Front-End stream (group and individual meetings) to provide clarity on tasks and help peers.
    5. Started upskilling on React.
    6. Figured out the problem with the article routes were BE related, hence notified the BE leader, Junkai to setup a trello card in the BE stream to fix this issue.
    7. No one in the FE team volunteered to become a coleader, so had to take on in managing the large FE team.
    8. Appointed two leads, Krish and Divanshi, to lead the sub group on the transition to React.
    9. Reviewed and merged the two commits from FE members on developing the reset password and register pages.
    10. Worked on improving a half built survey by Junkai, by adding new questions to it and rephrasing some of the existing questions. Then integrated it onto our webapp. Added it to the dashboard page.
    11. Met with Denica, the DS/AI stream leader, and discussed the updates that need to be in the dashboard.
    12. Reviewed and merged the feature to delete a user account to the develop branch developed by a FE member.
    13. The task on developing a feature for the user to leave a review was handed over to Saad Ali as he was looking for a task.
    14. Successfully completed developing an in-app notification system.
    15. Updated the BE leader that this task needs a corresponding BE task.
    16. Updated the dashboard by removing the title drop shadow, which was going out of the screen size, removing the three pie charts with dummy data and the two line graphs at the bottom of the page.
    17. Assigned the visualisation task to Gunjan Sharma and Sushant Dudeja. Gunjan was given the transaction data to be visualised and Sushant was given the home loan data to be visualised (2 graphs each).
    18. Reviewed and merged the work of the transition to React.
    19. Met with the Dolfin Leadership team to discuss and decide on the final responsibilities, the handover documents, and the upcoming deadlines.
    | +| Sri Harsha Vardhan |
  • Improved the previous BERT model with adding more epochs, dividing the input dataset into batches and also using attention masks on a new dataset to improve the accuracy of the model to 79%.
  • Also written code to save the best performing model, so that this model could be called upon to generate sentiment for a feedback text in the feedback form API.
  • Used mockaroo to generate fake feedback form data to generate graphs for the dashboard from this data.
  • Modified the visualization code for the feedback dashboard to also accept html files and not just the image files for the pie charts. This enabled me to fix the Sankey diagram error and display it in the frontend for the data generated as mentioned above.
  • Added code to the feedback API in the backend to store the data in a database table instead of a csv file and also added the function to call the sentiment analysis model generated by the BERT model, so that the review and it’s sentiment can be stored at the same time. This is not yet enabled as changes are being made to the database.
  • Also added sentiment analysis from textblob as a fallback because the 411MB model generated from the training couldn’t be pushed to github without complicating the repo by adding github large file storage.
  • Analysed all the savings prediction models and used the best performing random forest model to create a simplified version of it and pushed it to github.
  • Made changes in the frontend HTML files like the header and footer section, so that they contain the buttons for feedback dashboard and feedback form.
  • Also added wordclouds to the feedback dashboard for the review text from feedback form.
    • +| Gimsara Kenula Elgiriyage |
  • Met with the Product Owner and mentor weekly to update and discuss the product and tasks.
  • Created new Trello cards for Front End stream.
  • Assigned tasks to FE team members.
  • Hosted meetings with the Front-End stream (group and individual meetings) to provide clarity on tasks and help peers.
  • Started upskilling on React.
  • Figured out the problem with the article routes were BE related, hence notified the BE leader, Junkai to setup a trello card in the BE stream to fix this issue.
  • No one in the FE team volunteered to become a coleader, so had to take on in managing the large FE team.
  • Appointed two leads, Krish and Divanshi, to lead the sub group on the transition to React.
  • Reviewed and merged the two commits from FE members on developing the reset password and register pages.
  • Worked on improving a half built survey by Junkai, by adding new questions to it and rephrasing some of the existing questions. Then integrated it onto our webapp. Added it to the dashboard page.
  • Met with Denica, the DS/AI stream leader, and discussed the updates that need to be in the dashboard.
  • Reviewed and merged the feature to delete a user account to the develop branch developed by a FE member.
  • The task on developing a feature for the user to leave a review was handed over to Saad Ali as he was looking for a task.
  • Successfully completed developing an in-app notification system.
  • Updated the BE leader that this task needs a corresponding BE task.
  • Updated the dashboard by removing the title drop shadow, which was going out of the screen size, removing the three pie charts with dummy data and the two line graphs at the bottom of the page.
  • Assigned the visualisation task to Gunjan Sharma and Sushant Dudeja. Gunjan was given the transaction data to be visualised and Sushant was given the home loan data to be visualised (2 graphs each).
  • Reviewed and merged the work of the transition to React.
  • Met with the Dolfin Leadership team to discuss and decide on the final responsibilities, the handover documents, and the upcoming deadlines.
    • |