In the past decade, the emergence of Quantum Machine Learning (QML) at the crossroads of quantum computing and machine learning has led to significant advancements. In many fields accurate time series forecasting, a crucial element in critical decision-making across fields such as finance, climate science, energy, has traditionally been addressed using statistical and deep learning solutions. Despite the theoretical potential of quantum computers to offer exponential speedup for certain classes of problems, the practical execution of quantum circuits with a large number of qubits and extensive circuit depth faces challenges on current Noisy Intermediate-Scale Quantum (NISQ) devices due to the lack of effective quantum error correction. This paper presents a comprehensive overview of various QML architectures proposed in the literature to tackle this problem. A particular emphasis is placed on Variational Quantum Algorithms (VQA), considering the unique advantages they offer in the NISQ era, where hybrid classical-quantum approaches exhibit optimal quantum advantages. Our analysis covers different existing methods, discussing their advantages, highlighting their limitations, and providing a glimpse into future research directions.