Determining increments for aggregated time series data to determine impact of individual features

I'm working with a data source that provides itemised transactions, which I am aggregating into 1 hour blocks to determine a 'rate per hour' as the dependent or target variable - i.e. like a time series.

So far I've looked at Logistic Regression, Random Forest Regressor and Gradient Boosting Regressor and got reasonable results - but am really trying to determine the weighting/ impact of the independent variables, to see which have the biggest impact on the DV.

Would there be any value in decreasing the granularity of the aggregation from 1 hour to 30 or even 15 mins?

If it helps, the data is showing daily and annual seasonality.