In citation891/MCAC: Multivariate Chi-Square Anomaly Classification

knitr::opts_chunk$set(echo = TRUE)

Initial Take

It took a little bit of work to understand what was going on with the package; however, there was sufficient documentation within the package to start working through the provided functionality.

A few areas of concern/suggestions

There were quite a few errors/warning that I was getting such as, "You have removed the last 4 observations from the data set." and "Some variables contain more than 50 levels. Only the 10 most popular levels of these variables will be tabulated." and "cols( sourceAddress = col_character(), destinationAddress = col_character(), transportProtocol = col_character(), bytesIn = col_integer(), bytesOut = col_integer(), categoryOutcome = col_character(), ad.SCN = col_character(), IP_Pair = col_character(), Device_Name = col_character(), TIME_START = col_datetime(format = "")" None of these errors seemed to impact my ability to work with the dataset/package.

There is a data.R file which does not appear to be used. It seems like additional documentation which was added; however, you do have a similar file in the /man folder which provides similar information.

(minor concern) I see that the runMCAC calls for the packages to be loaded; however, if I am going into the /inst/apps/MCACapp and select the server.R or ui.R and try to run the app that way, it doesn't install/load the readr package. There is a nice error that clued me in on it though.

There is a sampleData.csv file and a sampleData.rda file. If the .rda file is selected for the shiny app, the shiny does not work. (If you read the directions on the shiny app, you would not make this mistake of course!) Unsure what the .rda file is used for if we're just using the .csv file.

Evaluation against intended targets

Data Upload Yes, this appears to have been accomplished. The Shiny app allowed for upload directly to the app.

Automatic Data Cleaning Yes, this appeared to work to clean the data using the prepareData and removeAnomaly functions

Automatic Time Vector Yes, it appears to be working. I think a little more description on this feature would be useful to allow for additional feedback.

Classify Outliers Yes, it appears that I was able to classify outliers using th RShiny App.

Generate Plot Yes, I was able to generate plots including the chi-square and Q-Q pltos within the Shiny App

Export Yes, this appeared to work; however, it does not seem that all the original attributes provided in the SampleData.csv file are provided ont he outliers.csv file.

Manual threshold I was able to move the slider and results seemed to change as the slider moved. It took a bit for it to recalculate.

Overall, it is a nice package.