I would like to analyse time series of RS derived map by R statistic. Does anyone know the best approach? I know stl and decompose, are they useful only for continuous data. What is necessary to do when data have some lack in the time series?
Ciro, "stl" with decomposition is definitely a good one. They are definitely useful for continuous data. The use of which time-series modelling approach also depend on what specific hypothesis you want to test, or on you research goals. These models can range from GLMs, GAMs ... to ARIMA models. The choice is yours! When data lacks some values, you could consider those values as missing at random.. If many of the values are missing, then the analysis could be confounded - leading to bias. I hope this helps!
I don't know what you mean by "RS derived map" and by "continuous data", but if your time series have missing data (denoted by NA), R arima can accept these, see Ripley Brian D., Time series in R 1.5.0, R News 2/2, June 2002, 2-7. I have checked the results. They correspond to what is obtained by the best interpolation algorithms (based on the Kalman filter and the fixed point smoother). Good luck!
Classical book which could be of interest is : Statistical and data analysis in geology, JC Davis - 1986 - J. Wiley. Deals with time series with "holes" in the data (among many other subjects).
Dear Ciro Manzo,
I am Ciro Manzo. So I have the same your first and last name !
This is very attractive and unusual for me......
Best wishes .
Dear Ciro
Have you done your time series analysis?
I propose using forecast package in R which has some good and strong tools to predict your time series.
- Badges
- Science topic
- Similar topics
- Statistics
- Time Series