Using MCrunch to analyze collected data


I have collected time-series data and would like to analyze it using MCrunch. I was hoping someone could give me some direction on how best to proceed. I would prefer to use the MCrunch settings file since I have been using it for previous FAST runs. However, I am running into problems when trying to load the data set.

To start, I have just one question.
What file formats can the settings file read? (.mat, .txt, workspace variable, etc) Or does the file extension need to be .out?

Or is there a different approach I should take when analyzing collected data? Such as modify MCrunch source m-files to perform stats on the data set, thus bypassing the settings file.



The nice thing about MCrunch is that you can do things pretty much your way. Once you load the data into the main structure, you can call whatever tools you like. You can create the main structure by using the parameter file and ReadManyFiles.m or write your own script. I did just that for CART2 test data.

MCrunch does not care what file extensions you use as long as the extensions of the input files do not conflict with the ones used for its output files. For instance, don’t put the data you want to analyze into a file called MyData.stat, because it will be overwritten when you compute the statistics.

Using the settings file requires the format of the data files to be formatted (human readable) and composed of a heading section with optional column names and units, followed by rows of columnar data. The data section must be numbers only. You can’t have columns with text in them to hold things such as the date in, say, a “Nov. 19, 2009” format. If you have that sort of thing, you should copy ReadManyFiles.m and modify it for your needs.

To save effort, I recommend you either use MCrunch as designed or modify ReadManyFiles.m to load unusual-format data into the FileInfo structure and call MCrunch’s analysis routines.