I have a data table which contains some discontinuation for the time series analysis. This is a monthly data table of the ocean water quality parameters for 15 years of data. Which imputation method should I use for the time series analysis?
Kalman or interpolation, Last Observation Carried Forward, or weighted moving average.
How to decide?
You set up a neural network to fill in the gaps, then import each of your variable as a column vector.
Depending on the choice of parameters there will be less or more differences between the monthly values. Any mentioned methods can be followed, try all three for comparison for a few discontinuations. Check some information from the World Ocean Database Select and choose "measured variables" (https://www.ncei.noaa.gov/access/world-ocean-database-select/dbsearch.html).
- Badges
- Science topic
- Similar topics
- Psychology