learn: Learn an nfer specification from an event trace.

In nfer: Event Stream Abstraction using Interval Logic

Learn an nfer specification from an event trace.

Description

This function applies the nfer mining algorithm to a trace to attempt to learn a specification. It will only learn before rules and will only find those when there is periodic behavior in the trace. This works best for system traces of real-time systems where events represent scheduled, periodic behavior. At this time, only the event names will be considered while any data is ignored.

Usage

learn(events, loglevel = 0)

Arguments

events	The dataframe containing events.
loglevel	(Optional) The logging level to set (0-4), where 0 is only warnings and errors and 3 is debug. Default is 0.

Details

Event traces are passed as dataframes with at least two columns. The first two columns contain event names and timestamps, respectively. Names should be strings. Timestamps should be integers or strings, ideally, but may be numeric. Subsequent columns will be ignored by this function, for now.

Example dataframe events:

| name | timestamp | x |    y |
-------------------------------
| foo  | 123       | 2 | NA   |
| bar  | 987       | 3 | TRUE |

Value

A handle to a learned nfer specification, loaded into R.

Examples

events <- nfer::read(system.file("extdata", "qnx.events", package = "nfer"))
learned <- nfer::learn(events)
learned <- nfer::learn(events, loglevel=1)

nfer documentation built on April 12, 2023, 12:40 p.m.