✳flâneur — a map of the web's best reading

Franklin Pezzuti Dyer

franklin.dyer.me · 3,967 words · saved by 1 readers

This is the story of how I wrote an absurdly inefficient algorithm to solve a cool problem in R, learned about copy-on-write semantics, and then came up with something much faster. This debacle has definitely made me a better R programmer. At my job, I’m working with a cyber intrusion dataset from the Canadian Institute for Cybersecurity that contains large lists of flows with associated statistics summarizing the traffic comprising each flow. These include stats like the flow’s total duration, the average rates of data transfer in packets per second and bytes per second, the mean and standard deviation length of a packet, and so on. A "flow" in this data set is confined to one of three network protocols (TCP, UDP and ICMP) and is identified by a pair of IP addresses identifying the source and destination, a pair of port numbers being used by the source and destination machines respectively, and a window of time distinguishing the flow from earlier and later flows. It happens that for

Discovering copy-on-write in R Discovering copy-on-write in R This is the story of how I wrote an absurdly inefficient algorithm to solve a cool problem in R, learned about copy-on-write semantics, and then came up with something much faster. This debacle has definitely made me a better R programmer. An aggregation problem At my job, I’m working with a cyber intrusion dataset from the Canadian Institute for Cybersecurity that contains large lists of flows with associated statistics summarizing the traffic comprising each flow. These include stats like the flow’s total duration, the average rat

Explore this link on the map →

related reading