I have M traces containing values of the received power. The duration of the trace is 5 seconds and contains 5000 samples. I want to do a statistical analysis of the small scale fading of this data using the windowing method. I would like to know the adequate length of the window to have meaningful statistical results. Does anyone know the right method to do it?
Article Sample size and power considerations for ordinary least squa...
Article Simulation-based power and sample size calculation for desig...
https://www.statisticssolutions.com/sample-size-determination/
https://stats.stackexchange.com/questions/257529/sample-size-calculation-for-interrupted-time-series
https://escholarship.umassmed.edu/qhs_pp/1292/
It depends on what the analysis is needed for.
I think in order to choose the best model for forecasting, you need to try different window lengths, carry out out-of-sample accuracy comparison and choose the optimal window length.
- Badges
- Science topic
- Similar topics
- Statistics
- Time Series