I have some realtime server data with total of 17 feature(column) in my dataset

question

#1

Is their any restriction on number of feature that i can provide to the htm anamoly algorithm.
Because most of the example i saw have maximum of three features.Thanks in advance for the help


#2

when you say features do you mean inputs or the number of variants your data can express it self with?


#3

i mean the column of my dataset like (disk_readcount,disk_writecount…etc)


#4

oh in that case YES

having more than one inputs gives you more space to find more patterns that are tied to other variables and hence giving you better and precise anomaly scores, depending on the type of data and other circumstances it may or may not take you more time to get some decent anomaly scores


#5

how can i provide multiple column as an input to the model .i have seen various examples but every example takes in only one column as an input .i did not find any example that provides multiple column as an input to the model.


#6

I think the best search on this topic is “multiple fields”.