Implementing mapWithState is the most suitable technique for ensuring stateful computation across batches in a Spark streaming application processing events from Azure Event Hubs, especially for windowed aggregations. Here's why:

1. Stateful Computation: mapWithState allows for maintaining arbitrary state information across batches, crucial for windowed aggregations where state tracking over time is necessary.

2. Efficient State Updates: It offers a more efficient way to update and manage state compared to alternatives like updateStateByKey, by allowing the definition of functions to update state based on incoming data.

3. Windowed Aggregations: It is particularly well-suited for windowed aggregations, enabling stateful computations within specific time windows, which is essential for calculating aggregations over fixed intervals in streaming data.

4. Scalability: Designed for large-scale stateful computations in distributed environments like Spark streaming, mapWithState ensures scalability and fault tolerance.

5. Real-time Processing: It facilitates real-time processing of streaming data from Azure Event Hubs while maintaining stateful computations across batches, allowing for continuous updates to aggregations as new data arrives.

In summary, mapWithState provides an efficient, scalable, and reliable method for stateful computation in Spark streaming applications, making it the optimal choice for processing events from Azure Event Hubs with windowed aggregations.