NLP, Neural Networks, Transformers, LLMs, Data Visualisation/Transformation, PostgreSQL and other ML/DS approaches.
Featured Project
Fake reviews detection.
The aim of this project is to detect fake reviews. I began by exploring traditional machine learning algorithms, including Random Forest, Naive Bayes, and Logistic Regression. These algorithms didn't give me the desired result, hence the need to explore advanced methods such as transformers, recurrent neural networks, and large language models (LLMs). After thorough evaluation, the XLNet transformer had the best performance with a 96% test-accuracy and a 94%- train accuracy. .
The model was trained on this dataset and deployed using FastApi and Docker.
UK BEIS SPENDINGS
This is a data visualisation project (using R markdown's - Flexdashboard ) which focuses mainly on Energy spendings by the Department for Business, Energy and Industrial Strategy (BEIS), UK. In this project, I collated 3 months worth of data, handled missing values, errors and duplicates. Then preprocessed the dataset to filter energy transactions.
Spendings over 1 million , 10 million and 100 million GBP were highlighted with respect to suppliers, supplier type, date of payment, amount and expense area.
Finally, different types of plots were used to visualise this data, showing the ease of use and versatility of Flexdashboard and R.
CLICK HERE TO VIEW THE DASHBOARD ONLY
Advanced Store Registration Using Postgresql Procedures and Functions
This project enables store owners to register new customers for their store. Additionally, it allows existing customers to make purchases by specifying the products they want, along with their preferred delivery date and time. The system ensures that delivery can be booked only if the product and delivery slots are available.
