I was working on a model and it seems from the plot that it might have outliers so i decided to use the tsouliers package to detect it. i have a couple of questions related to that package
- the tso function of the tsoutliers package locate the outliers and and shows a potential outliers you might have but i noticed that from the plot that my data might have a lot more than that, is that a statistical problem from the function or it might lack the functionality ?
- the arima model that it suggest , is it the final suggested fit after considering the detected outliers or is it the one was primary or initial model it used to detect them?
- I also learnt about tsoutliers function which detect and suggest a replacement for the detected outliers, but when i used it on a data set, the results were 0 for all while i used tso function and it already detect an outliers , so how is that possible? do they work differently , or do i have to change some perimeters ?
- from experts with it view , do you think its a powerful method to detect and deal with outliers or it still have issues and lack the functionality ?
- please feel free to share tsoutliers or outliers detection in time series resources other than the tsoutliers documentation (like real life data examples) so i could get a better understanding of its methods