Machine Learning-Based Stock Market Behaviour Analysis Using K-Means Clustering

Stock market behaviour involves complex interactions among various indicators, like market trends and diverse data. Traditional analytical tools often struggle to capture this complexity, especially with long-term relationships and multiple data sources. Machine learning offers a formal way to analyse these complexities by revealing hidden structures and key indicators that drive market dynamics. Systematically integrating different data sources can boost predictive reliability and support smart investments. A machine learning model, using clustering and deep learning, examined stock market behaviour. K-Means clustering grouped similar market patterns, structuring the dataset into meaningful clusters. To consider long-term dependencies in financial information, a Long Short-Term Memory (LSTM) network was used. An attention feature was added to the LSTM framework to select signals that influence movements over time. This hybrid LSTM–Attention framework enabled effective integration of various data sources and focused on relevant features affecting stock market movement. This hybrid approach identifies significant market patterns in a structured way, enhancing predictive accuracy and using multiple data streams to support data-driven investment and portfolio creation.