This study introduces a novel method to enhance predictive model performance for estimating continuous well production flow rates using daily operational conditions. The approach leverages machine learning algorithms, employing bagging as ensemble technique, to consolidate predictions from multiple base models. The algorithms are trained and evaluated using operational data, including wellhead pressure, temperature, choke diameter, gas lift injection rate, and gas-oil ratio (GOR), among others. The predictive model is evaluated using historical data from a mature offshore field, where GOR increases over time. Results demonstrate the ensemble machine learning model's high accuracy, surpassing traditional Gilbert-type choke correlations, which are constrained by the applicable range of operations variables like GOR and choke diameter. These findings suggest that this approach provides a reliable and efficient solution for estimating well production flow rates, particularly when continuous flow rate measurements face limitations.