You are tasked with developing a machine learning model to predict house prices for a real estate company. During the data preprocessing phase, you discover that the 'distance from the nearest school' feature, which is considered a crucial predictor, has a significant number of missing values and exhibits low variance. The company emphasizes the importance of utilizing every data row to maximize the model's predictive accuracy. Additionally, the solution must be cost-effective and scalable to accommodate future data. Given these constraints, what is the optimal strategy to handle the missing data in this scenario? Choose the best option.