Revealing User Behavior Trends through Time-series Analysis of Google Analytics Data

Nowadays, in the digital era, data has become an invaluable resource for organizations. The analysis of this data is crucial for strategic decision-making and business process optimization. In this context, Google Analytics emerges as an essential tool, providing detailed insight into online user behavior. This study applies advanced time series techniques using three different approaches: traditional statistical models such as ARIMA, LSTM neural networks, and a third model based on LSTM with hyperparameter optimization. Imolko is dedicated to offering purposeful email marketing services, where the data analyzed is a collection of events performed on the website, classified into events by category and events by action, this data spans a period from May 2020 to May 2021 whose objective is to identify behavioral patterns, understand trends over time and explore correlations. The ultimate purpose is to provide valuable insights that drive strategic decision-making and contribute to the organization's sustainable competitive advantage in a dynamic business environment. The study found that LSTM models, particularly those with hyperparameter tuning, significantly outperformed traditional ARIMA models in predicting user behavior, achieving an accuracy of 63.6%. The analysis highlighted the importance of understanding customer interactions to optimize marketing strategies and improve customer satisfaction. Overall, the findings emphasize the importance of advanced neural network architectures and hyperparameter tuning in achieving precise time series predictions, offering valuable insights for strategic decision-making in digital business environments.