Projects

The project aims to predict the productivity of garment employees using the dataset from the UCI Machine Learning Repository. It highlights the importance of tracking, analyzing, and predicting productivity in the garment industry. The dataset includes various features such as date, quarter, department, day, team, targeted productivity, and actual productivity, among others. Initial observations note issues like actual productivity values surpassing the limit of 1, a specific number for work in progress, and efficient time management with minimal idle time and idle men. Further details

The multinomial Naive Bayes algorithm is used along with a dataset of 5,572 SMS messages that are already classified by humans to build a spam filter

This project is about building a spam filter for SMS messages using the multinomial Naive Bayes algorithm. The goal is to classify messages as either spam or ham (non-spam). Further details

This ML project is focused on building a classifier to predict heart disease based on various features. The notebook starts with data exploration and preprocessing, including handling missing values and feature selection. It then proceeds to build a k-Nearest Neighbors (k-NN) classifier using different features and hyperparameter optimization. Further details

This project develops a TensorFlow-based deep learning model to predict listing gains for IPOs in the Indian market. Listing gains refer to the percentage increase in the share price from its IPO issue price on the listing day. This model aims to assist an investment firm in making informed decisions about investing in IPOs. Further details

n this project, I imported a dataset of dog images, train a convolutional neural network to classify those images, and improve the model performance. The original data is from Stanford, and it contains images of many dog breeds. In this project, however, I use a reduced dataset containing only five breeds. Further details

The UCLA University Admission Data Set was evaluated to see if the chance of admission can be predicated. The aim of the project is to learn how to optimise machine learning hyperparameters using the following three strategies:

• Grid Search
• Randomized Search
• Bayesian Optimization

Further details

A Tableau dashboard was developed through a few iterations to evaluate the booking data for workshops run by the digital skills lab at LSE. This included preparing the data through Tableau Prep. Further details

The aim of the project is to understand what might influence the success of a campaign on Kickstarter. Further details

The goal of this project is to analyze data from a sales records database for scale model cars and extract information for decision-making. Further details

This project aims to provide valuable insights to the management of Northwind Traders by leveraging PostgreSQL window functions on the Northwind database. The focus areas include:

• Evaluating Employee Performance: Recognizing top-performing employees and identifying those needing support.
• Understanding Product Sales and Category Performance: Optimizing inventory and marketing strategies.
• Analyzing Sales Growth: Identifying trends, monitoring progress, and making accurate forecasts.
• Evaluating Customer Purchase Behavior: Targeting high-value customers with promotional incentives.

Further details

Implementation of 4-matrix educational data anlysis tool across Trust schools and Arbor MIS system

Design and implementation of sharepoint across secondary and 2 primary schools; this was part of the edtech strategy I wrote for the Trust. We had partnered with Cloud Design Box

Lead TIBHS flagship school at Star Academies to Microsoft Showcase status
Lead Trust schools with remote learning during COVID-19 lockdown

Implementation of Canvas LMS to facilitate online learning; this was part of the digital education strategy I wrote for the school

Implementation to Google Education to support self-paced learning with C3 coursework (MEI),
OneNote and GeoGebra for teaching A-level Mathematics