Your task is to choose the most appropriate data structure (list, ndarray, set, or tuple) for each of the different scenarios below. Based on the nature of the data and the specific requirements for each collection, fill in the blanks (___) to create the correct data structures. Make sure to use the appropriate brackets for each structure, and if creating a NumPy array, initialize it based on a list.

1. For monthly_sales, the data is numerical, and you will need to perform frequent calculations on it.
2. user_ids, each ID should be unique, and you will need to frequently check if a certain ID is present.
3. product should be a fixed (unchangeable) record.
4. tasks should be an ordered collection and allow for frequent additions and removals.