Filter:
02 / DEEP LEARNING
Mask Wearing Detection
Trained a CNN using TensorFlow to detect mask compliance from images. Applied data augmentation, dropout regularisation, and evaluated on real-world test sets for a public safety use case.
03 / NLP
Online Shopping Review Analysis
Sentiment analysis pipeline on e-commerce reviews. Tokenisation, stopword removal, TF-IDF vectorisation, and classifier training — surfacing key product pain points for non-technical audiences.
04 / MACHINE LEARNING
Insurance Quote Prediction
Predictive model built in KNIME to estimate insurance quotes from customer attributes. Focused on decision tree interpretability and model explainability to support pricing strategy.
05 / ALGORITHMS
Decision Tree from Scratch
Full implementation of a decision tree classifier in pure Python — recursive splitting, Gini impurity, tree construction, and prediction — without any ML libraries. Demonstrates deep algorithmic understanding.
06 / DATA ANALYTICS
Bellabeat Smart Device Analysis
Organised and cleaned 18 CSV files (322MB) of fitness tracker data. Visualised user behaviour trends across activity, sleep, and calories. Delivered 4 strategic marketing recommendations in a client-ready report.
07 / SQL
Covid-19 Global Data Exploration
SQL Server analysis of global Covid-19 data — death rates, infection rates, vaccination rollout by country and continent. Used CTEs, window functions, and temp tables to extract meaningful epidemiological insights.
08 / VISUALISATION
Covid-19 Dashboard
Interactive Tableau dashboard visualising global Covid-19 trends built on the SQL exploration dataset. Includes global death map, infection rate trends by country, and a death percentage forecasting chart.
09 / DATA EXPLORATION
Movie Industry Exploration
Python analysis of movie industry data covering budget, gross revenue, ratings, and genres. Performed data cleaning, outlier detection, and correlation analysis to identify what factors drive box office success.
10 / SQL
Nashville Housing Data Cleaning
SQL-based data cleaning on Nashville housing dataset — standardised date formats, populated missing property addresses via self-joins, split address columns, removed duplicates with ROW_NUMBER(), and restructured data for downstream visualisation.
11 / VISUALISATION
Nashville Housing Dashboard
Tableau dashboard built on the cleaned Nashville housing dataset. Visualises property sale trends by year, land use breakdown, owner vs vacant status, and geographic distribution across Nashville zip codes.
← Back to Home