List of my Sample Projects on Machine Learning
Click on Project Name to View the Project
- Predicting Real Estate Prices (Python): The real estate market in Sindian District, New Taipei City, Taiwan is influenced by multiple factors, including property age, accessibility to public transportation, neighborhood amenities, and geographic location. Determining an accurate property value can be challenging because these factors interact in complex, non-linear ways. Traditional appraisal methods often rely on expert judgment and historical comparisons, which can be time-consuming, subjective, and may not fully capture changing market conditions.
- This project addresses the business need for a data-driven house price prediction model that enables more accurate and consistent property valuation. By leveraging machine learning techniques, the model uses key property attributes such as transaction date, house age, distance to the nearest Mass Rapid Transit (MRT) station, number of nearby convenience stores, and the property's latitude and longitude to estimate house prices.
- Prediction of Timely Payments (Python): Every business that invoices customers faces one common challenge: 1. Ensuring invoices are paid on time, and 2. Late payments can disrupt cash flow, increase administrative costs, and affect the company’s ability to operate efficiently. "We Provide All", a mid-sized B2B service provider, issues hundreds of invoices monthly to clients across various industries. Some clients pay promptly, while others consistently delay payments. The Operations Department want to anticipate payment delays so they can take preventive action such as early reminders, offering discounts, or adjusting payment terms.
- This project is a machine learning pipeline for predicting timely payments (arrears vs. on-time) using financial and behavioral indicators. It is aimed at improving the performance of "We Provide All" in terms improving ontime payments. The notebook demonstrates a complete end-to-end workflow including data preprocessing, model training, hyperparameter tuning, and evaluation.
- Predicting Credit Card Fraud (Python):Financial institutions and payment service providers face significant challenges in detecting and preventing fraudulent transactions. As digital payments continue to grow, fraudulent activities can result in substantial financial losses, regulatory risks, and reduced customer trust. Traditional fraud detection methods that rely on predefined rules may struggle to identify evolving fraud patterns, especially when dealing with large volumes of transaction data.
- This project addresses the business need for an automated fraud detection system by developing a machine learning model that can classify credit card transactions as either fraudulent or legitimate. Using Logistic Regression, the model analyzes transaction patterns from a dataset containing 30 features and 284,807 records, including anonymized variables generated through Principal Component Analysis (PCA), transaction amount, and transaction classification labels. The goal is to accurately identify suspicious transactions while minimizing incorrect classifications that may disrupt legitimate customer activities.
- Predicting Credit Card Fraud (Python):Financial institutions and payment service providers face significant challenges in detecting and preventing fraudulent transactions. As digital payments continue to grow, fraudulent activities can result in substantial financial losses, regulatory risks, and reduced customer trust. Traditional fraud detection methods that rely on predefined rules may struggle to identify evolving fraud patterns, especially when dealing with large volumes of transaction data.
- This project addresses the business need for an automated fraud detection system by developing a machine learning model that can classify credit card transactions as either fraudulent or legitimate. Using Logistic Regression, the model analyzes transaction patterns from a dataset containing 30 features and 284,807 records, including anonymized variables generated through Principal Component Analysis (PCA), transaction amount, and transaction classification labels. The goal is to accurately identify suspicious transactions while minimizing incorrect classifications that may disrupt legitimate customer activities.
- KMeans++ for Mall Customer Insight (Python): Understanding customer behavior and preferences is essential for organizations that want to design effective marketing strategies and improve customer engagement. However, customers often have diverse characteristics, purchasing behaviors, and spending patterns, making it challenging for businesses to apply a one-size-fits-all approach to marketing and promotions.
- This project addresses the business need for a customer segmentation solution by applying machine learning clustering techniques to group customers with similar characteristics. Using PySpark and the K-Means++ clustering algorithm, the model analyzes customer attributes such as gender, age, annual income, and spending score to identify distinct customer segments based on behavioral and demographic patterns.The resulting customer clusters can help the administration make more informed, data-driven decisions for marketing campaigns, promotional strategies, and customer relationship management. Instead of targeting all customers with the same offers, businesses can develop personalized approaches tailored to the needs and preferences of each customer group.
- Predicting High Risk Credit Card Applicants (R): Financial institutions face the ongoing challenge of evaluating credit card applications accurately while balancing business growth with financial risk management. Approving high-risk customers can lead to increased chances of payment defaults, financial losses, and higher collection costs, while overly strict approval decisions may limit opportunities to acquire valuable customers. Traditional credit assessment approaches may not always capture complex relationships between customer characteristics and credit risk.
- This project addresses the business need for an automated credit risk classification model that helps financial institutions identify potentially high-risk credit card applicants before approval. Using a Support Vector Machine (SVM) model, the system analyzes customer lifestyle variables and behavioral attributes to classify applicants as either high-risk or lower-risk credit card customers. The model provides an additional data-driven reference that can support existing credit evaluation processes and improve risk assessment decisions.
- KMeans++ for Mall Customer Insight (Python): Understanding customer behavior and preferences is essential for organizations that want to design effective marketing strategies and improve customer engagement. However, customers often have diverse characteristics, purchasing behaviors, and spending patterns, making it challenging for businesses to apply a one-size-fits-all approach to marketing and promotions.
- This project addresses the business need for a customer segmentation solution by applying machine learning clustering techniques to group customers with similar characteristics. Using PySpark and the K-Means++ clustering algorithm, the model analyzes customer attributes such as gender, age, annual income, and spending score to identify distinct customer segments based on behavioral and demographic patterns.The resulting customer clusters can help the administration make more informed, data-driven decisions for marketing campaigns, promotional strategies, and customer relationship management. Instead of targeting all customers with the same offers, businesses can develop personalized approaches tailored to the needs and preferences of each customer group.
- Predicting High Risk Credit Card Applicants (R): Financial institutions face the ongoing challenge of evaluating credit card applications accurately while balancing business growth with financial risk management. Approving high-risk customers can lead to increased chances of payment defaults, financial losses, and higher collection costs, while overly strict approval decisions may limit opportunities to acquire valuable customers. Traditional credit assessment approaches may not always capture complex relationships between customer characteristics and credit risk.
- This project addresses the business need for an automated credit risk classification model that helps financial institutions identify potentially high-risk credit card applicants before approval. Using a Support Vector Machine (SVM) model, the system analyzes customer lifestyle variables and behavioral attributes to classify applicants as either high-risk or lower-risk credit card customers. The model provides an additional data-driven reference that can support existing credit evaluation processes and improve risk assessment decisions.
- Classification of Exercise Movements (R): The increasing adoption of wearable fitness devices, motion sensors, and digital health applications has created a growing need for accurate and automated recognition of physical activities. Correctly identifying exercise movements is important for fitness tracking, rehabilitation programs, sports performance analysis, and health monitoring. However, manually analyzing movement patterns can be time-consuming, subjective, and difficult to scale across large numbers of users.
- This project addresses the business need for an automated exercise movement classification system that can accurately identify different types of physical activities using machine learning techniques. By analyzing movement data collected from exercise activities, the model aims to classify exercise patterns and determine the correct movement category with high accuracy.
- Classification of Iris Plants (Python Tensorflow): Accurately identifying plant species is essential in agriculture, botanical research, biodiversity conservation, and environmental monitoring. Traditional classification methods rely on manual inspection by experts, making the process time-consuming, labor-intensive, and susceptible to human error, particularly when dealing with large datasets or species with similar physical characteristics.
- This project addresses the need for an automated and reliable plant species classification system by developing a Multi-Layer Perceptron (MLP) machine learning model using the TensorFlow framework. The model analyzes four measurable flower characteristics sepal length, sepal width, petal length, and petal width to classify an iris flower into one of three species: Iris Setosa, Iris Versicolor, or Iris Virginica.
- Classification on Wine Data using KNN (Python): The wine industry relies heavily on accurate product classification to support quality control, inventory management, pricing strategies, and customer satisfaction. With a wide variety of wine products available, manually identifying and categorizing wines based on their chemical properties and characteristics can be challenging, time-consuming, and dependent on expert knowledge. Inconsistent classification may also affect product positioning, quality assessment, and consumer purchasing decisions.
- This project addresses the business need for an automated wine classification system by developing a machine learning model capable of identifying wine categories based on measurable characteristics. Using K-Nearest Neighbor (KNN) algorithms with different distance metrics, the model evaluates patterns within wine feature data to classify samples into their appropriate categories. The performance of each KNN approach is assessed based on both classification accuracy and computational efficiency, ensuring a balance between predictive performance and practical usability.
- Classification of Iris Plants (Python Tensorflow): Accurately identifying plant species is essential in agriculture, botanical research, biodiversity conservation, and environmental monitoring. Traditional classification methods rely on manual inspection by experts, making the process time-consuming, labor-intensive, and susceptible to human error, particularly when dealing with large datasets or species with similar physical characteristics.
- This project addresses the need for an automated and reliable plant species classification system by developing a Multi-Layer Perceptron (MLP) machine learning model using the TensorFlow framework. The model analyzes four measurable flower characteristics sepal length, sepal width, petal length, and petal width to classify an iris flower into one of three species: Iris Setosa, Iris Versicolor, or Iris Virginica.
