The Accounting Analytics Capstone course is a compulsory capstone course for undergraduate students pursuing a 2nd Major in Accounting Data & Analytics. We showcase their completed projects below.
Developing a Model to Make More Informed Decisions Related to Anomaly Detection
The team adopted a Three-Pronged approach for Data Preparation - Explicit Rules, K-Means Clustering and Random Forest, and combined the outputs to identify the common stocks flagged for anomalies across the selected period by all 3 models. They also developed a Tableau dashboard to provide a comprehensive analysis of each stock’s performance in terms of both selected market indicators and financial ratios.
Detecting Anomalies in Stock Data, Focusing on Companies Listed in the NASDAQ-100 index
The team utilised Machine Learning models like LongShort-Term Memory (LSTM) and ARIMA GARCH to predict stock price movement and built a Dashboard in PowerBI to present the analysis models, and to help with visualisation of the predictions.
Leveraging a Predictive Model to Estimate Box Office Performance to Aid Operations and Decision-Making
The team aimed to uncover the most influential factors (predictors) that drive box office performance in order to develop a predictive model to estimate box office for upcoming films, factoring in historical data and other relevant characteristics.
Box Office Forecasting & Movie Session Optimization
The team aimed to develop prediction models that give deeper insights to trends, forecast revenue through movie attributes and aids in optimising session scheduling. Their main approach involved identifying key factors affecting movie attendance, forecasting box office revenue and admissions using machine learning, and optimizing scheduling with goal
programming and data visualization.
Data Driven Movie Prediction & Optimisation Model
The team developed a Box Office Prediction Dashboard and a Session Optimisation Dashboard to help Cathay Cineplexes predict the performance of a prospective movie, and subsequently optimize the programming for the movie.
Automating the Portfolio Construction Process and Allocation of New assets to Existing Portfolios
The team developed a model to automate the typically manual portfolio construction process and efficiently allocate new assets to existing portfolios.
Making the Mean-Variance Optimisation (MVO) Model More Dynamic
The team aimed to identify the models that generated the highest market returns depending on the market state and established the following recommendations:
Stable Market: Sharpe Base
Sharp Market Decline: Sortino Base
Market Recovery: HMM SortinoMVO
Optimising Operations and Avoiding Inventory Stock Outs by Establishing a More Precise Reorder Point and Utilising Data Visualisation to Analyse Inventory Levels
The team developed a predictive model to establish reorder points and alerts, as well as a Dashboard that highlights current and forecasted product demand, aiding DrinkAid in planning future reorder quantities for each component and strategic decisions.
Utilising Revenue and Advertising Data Insights to Minimise Revenue Losses and Optimise Business Expenditure
DrinkAid's manual monitoring of financial metrics has resulted in delayed detection of critical Issues. The team had the following objectives:
- Automated data processing for future updates
- Automated dashboarding
- Anomaly detection
- Forecasting of Sales, and Return on Advertising Spending (ROAS)
