: Time series forecasting plays a vital role in data-driven decision-making across various domains. This thinking centers on the modeling and recreation of univariate and multivariate analytics utilizing profound learning and machine learning strategies. Other calculations are connected to foresee time arrangement information within the neural arrangement show, particularly for the Sylhet locale of Bangladesh. The investigation investigates distinctive estimating approaches, counting conventional machine learning models. The execution of these models is assessed utilizing key mistake measurements such as RMSE, R-squared, MAE, and MAPE to decide their precision and effectiveness. The discoveries provide experiences in the adequacy of distinctive strategies in capturing complex worldly designs in univariate and multivariate datasets. This considers points to improve prescient analytics for climate, financial matters, and other time-dependent components within the Sylhet locale, contributing to made strides in decision-making and key arranging.