# Forecasting with ARIMA models and sensor data in R

I'm trying to apply ARIMA models to sensor data and would be thankful if anyone could answer my questions. I should add that I have very little experience with time series (trying to change that).

The data concerns the spindle load and the spindle RPM on a CNC machine, collected at irregular intervals of 10 to 15 seconds. I downsampled the data to 5 minute intervals, not only to make it regular, but also because 5 minute forecasts are sufficient for my purposes (10 or 15 minutes might also work). After resampling, both time series have 755 observations.

My first question is: can I actually apply models like ARIMA, ETS, Holts Winter, etc to this data? I'm trying to do so because I read that ARIMA is often used to model sensor data, however I've also read that these models don't work very well with very high frequency data. The following are plots of the spindle load and spindle RPM, respectively:

I applied a Box Cox transformation (lambda = 0.1741072) to the spindle load data and differenced it to make it stationary and then used auto.arima to find an appropriate model. The spindle RPM data was already stationary. In both cases, the residuals appear to be white noise, but how can I be sure these are actually good models for forecasting?

Question 2: what frequency should I choose when creating ts objects in R? Since I was having some trouble defining this parameter, I decided to use the findfrequency function from the forecast package which returned 1 for the spindle load (so, no strong periodicity) and 10 for the spindle RPM. I understand this corresponds to the maximum of the spectral density, but I'm unsure on how to interpret 10 in the time domain. Does it correspond to a period of 2.4 hours?