UBA | UEBA Solution for Insider Security

You probably want //*[contains(concat( " ", @class, " " ), concat( " ", "listaLinks", " " ))]//a/@href

You missed modifying the value post_max_size in the php.ini file. Change it to 950M to have posts accept the larger memory buffer. Restart Apache and the new value will take effect.

For now data is a JSON content : a string, you need to load it as a JS structure, here a list of objects

Parse the JSON, and also use data[i].total_amount and not data.total_amount

I think this does what you ask. Notice that I have re-organized the code to put all of the input stuff in front, so that the p/u dependent stuff is at the end. Note that I may have misunderstood the relationship between the size of the A and C matrices and the size of the B matrix, so check this carefully.

Note that numpy includes a transpose routine. You don't need to do it by hand. I haven't decomposed the computation of Xab amd X to see if those could use numpy primitives, but I wouldn't be surprised.

Try adding some sleep time (say 3 seconds) every n number of tickers.

One way is to write a simple function, that takes in a vector of packages and suppresses the output in a way you want, as examplified below.

After spending about an hour, thanks to @user20650 and @GWD I was able to solve my problem as follows:

• I previously had R 3.5.3, I upgraded to the latest version 3.6.2
• Steps to upgrade R if you already have it: tutorial 1 and tutorial 2
• After having R 3.6.2, I used RGui rather that RStudio to install Hmisc
• I typed in the GUI console: install.packages("Hmisc"); it prompts you for a message about compilation, I clicked on NO

It looks like you have a couple relatively large values, while most are small. This causes 'skewing' of your color scale in a way that looks less interesting. Instead of graphing number of users, consider mapping based on a quantile .

Here is an implementation of grouping into quantiles using cut2() from Hmisc. In this case, values are cut into 5 groups.