A weather forecasting system typically consists of three modules.

Data Collection: Real$-$time data $($i$.$e$.,$ data is being collected continuously$)$ is gathered from various sources.

Data Analysis: The collected data is analyzed using different mathematical models and algorithms.

Future Predictions: Forecasts for future weather conditions are generated based on the analyzed data.

However, the initial step is to efficiently store the weather information $($e$.$g$.,$ location, temperature, humidity, and wind speed etc.$)$ in memory so that it could be conveniently used in numerous modules having various aspects. Being a Java developer, your task is to decide whether to use primitive data types to manage the data efficiently, or to create custom objects $($i$.$e$.,$ instances of classes$)$ to fulfil the need of various modules in a typical weather forecast system.

Question:

Keeping Java$$s memory management practices in view, which option $($i$.$e$.,$ either primitive data types or custom objects$)$ will you adopt for the above given situation, and why? Support your choice with solid reasons.