In this tutorial we will learn how to run sift on the command line, loading a library, running queries then running PCA and using that for outlier detection. once we run through the commands step by step we will use them to create a batch script that can be used to execute them automatically, we will include so logic to automate the process even further.

1. First we need to run Sift on the command line, on your keyboard press the Windows Key + R in the run window type in cmd and press enter, this will open a command line terminal.
   
2. Then change the directory to Sift's install location
   
3. Next to launch sift we need to call the .exe and pass the argument –NoGUI
   
4. This will open another console that we can use to interact with Sift
5. In the new console we will load a library, enter the command –LoadLib directory “Path to downloaded library”
   
6. Next load and run a predefined query by entering the command –LoadQuery file “Path to query definition file”
   
7. Select all signals by running –SelectSignals (Ensure all signal groups and work-spaces have been selected)
   
8. Run a PCA with the name “Outlier_PCA” –RunPCA name “Outlier_PCA”
   
9. Now we will run our first round of outlier detection run an Local Outlier Factor test with the command –RunLOF grouping “group” autoExclude This will find any group level outliers and exclude them automatically
   
10. Now lets export the results of this LOF test run the command: –ExportPCA file “Path to exported txt file” exportFormat “transposed” LOF LOFThrehsold
    
11. Next we will run another LOF test, this time on workspaces: –RunLOF grouping “workspace” autoExclude
    
12. Now export the results of this LOF test as well: –ExportPCA file “Path to exported txt file” exportFormat “transposed” LOF LOFThreshold
    
13. Finally, lets save the project so we can view it in the GUI if needed: –SaveProject file “Path to saved project”
    

Now our processing is complete, we have two exports detailing group and workspace level outliers and a saved project file that has excluded all of the found outliers. We can use this as a baseline for future processing.

Now that we have an understanding of what we want as a baseline for processing, we can take the steps to automate it!

In order to fully understand how we should automate a process we need to understand all the details, so lets create a scenario.
Lets say we are working with a baseball team, and every day at practice we get some data that needs to be processed, that data will be stored in dated folders in a root directory something like this:
Within each of these folders are three trials which contain relevant C3Ds

Now that we know what we are working with we can think of how to build our batch script, first we will need to check each folder within the root directory, and if the folder does not already contain our exports (.txt files and .i3d) then we know this folder needs to be processed.

@echo off
setlocal

set "targetDir=C:\Users\shane\OneDrive\Documents\practice"

for /D %%d in ("%targetDir%\*") do (
    set "i3dFileFound="
    for %%f in ("%%d\*.i3d") do (
        if exist "%%f" (
            set "i3dFileFound=true"
        )
    )
    
    if not defined i3dFileFound (
        set "folderWithoutI3d=%%d"
        "C:\Program Files (x86)\Sift\Sift.exe" --NoGui --LoadLib directory "%%d" --LoadQuery file "C:\Users\shane\OneDrive\Documents\practice\query_definition.q3d" --selectSignals --RunPCA name "Outlier_PCA" --RunLOF grouping "group" autoExclude --ExportPCA file "%%d\Group Outliers.txt" exportFormat "transposed" lof lofThreshold --RunLOF grouping "workspace" autoExclude --ExportPCA file "%%d\Workspace Outliers.txt" exportFormat "transposed" lof lofThreshold --SaveProject file "%%d\Outliers.i3d" --Exit
    )
)

endlocal