Barclays Problem Statements


PRODIGY9: Problem Statement Sharing Session

Barclays: Problem Statement - Support Mail/BOT Prediction

Prize: Lunch & Networking/Interview Session with Barclays Singapore CIO

1st place: $450 USD - Microsoft Funded
2nd place $225 USD - Microsoft Funded

Every application has Support Mailbox that receives numerous emails in a day from users requiring assistance. We have resources monitoring the emails on shift basis across the globe. Some of the emails may be recurring and may be pointing to knowledge bases or known resolutions. The idea is to filter and categorize these mails where knowledge base is already available, or an informed action can be taken without any manual intervention required thereby reducing the efforts spent by resource on these emails.

Problem Statement:
Design an Algorithm that will perform data analytics on a given email data and categorize email data belonging to specific category and predict the relevance. The algorithm should also filter out the spams in addition to the category that has been filtered out earlier. Some of the Categorization can be Resumes, Application, Meetings, HR, Spam, etc.

Data Set:

Data Set Tips:
Use email body for the spam detection and since the dataset has already has minimum classification done for the emails use the folder as a feature for classification.
1) Explore the data. (Visualization, Standardization)- Provide your interpretation of each step.
2) Data Preprocessing and cleaning (Can use Bag of words model when dealing with text)- Detailed explanation is required for each step.
3) Model Building- Use Logistic Regression
4) Categorization/Prediction and Performance Evaluation- Interpret the prediction accuracy.

Tech Tips:
SK learn , Logistic Regression.

1. Please follow the naming convention as _. ipynb
2. Inside each jupyter notebook, you are required to mention your name, Group details and the Assignment dataset you will be working on.
3. Organize your code in separate sections for each task. Add comments to make the code readable.
4. Notebooks without output shall not be considered for evaluation.
6. Delete unnecessary error messages and long outputs.
7. Display the analysis of attributes in one frame rather than one after one. However, special treatment to attributes can be displayed separately.
8. Prepare a jupyter notebook to build, train and evaluate a Machine Learning model on the given dataset.
9. Only two files should be uploaded without zipping them. One is ipynb file and other on html output of the ipynb file. No other files should be uploaded.

Tech Stack:
Azure ML

PRODIGY9: Problem Statement 2 - Create an AI service to compose a new character

Prize: 3-month internship at PRODIGY9 Co., Ltd. (Bangkok or Remote) and a special CWB edition custom designed scarf

Given a large image dataset from character making websites such as or which contains images that can be composed together to make a character (for example, swords, shields, helmets for a knight).

We would like you to create an AI service such that when given a prompt, it would select the individual parts from the library of images at your disposal to compose a new character that would best match the given prompt. If no images are available or the AI believes there are not enough data/images available to satisfy the prompt, we would like for it to simply make a suggestion that will most closely match the prompt.A small set of images will be provided for testing.

Tech Stack:
– Azure AI Vision
– Azure OpenAI

Success Criteria:
– Creativity on how the prompt UI is presented.
– Quality of the generated result.
– AI usage costs per prompt

Sponsored by:

Powered by:

Stay connected: