It is a tool to help with initial exploratory data analysis on the command line. It is meant to be used together with all the other great tools available.
Specifically, automat provides following functions to help you wrangle with your data:
- filter (can filter numerical values)
- mutate
- summarize
- arrange
- group_by
- other, SQL-like operations on your tabular data
The csv used in all the examples and benchmarks is the worldcitiespop dataset from the Data Science Toolkit.
Simple filtering:
atm worldcitiespop.csv filter "Population<1000000"
Multiple filter commands can be chained together:
atm worldcitiespop.csv filter "Population<1000000"|atm filter "Longitude<-50"
atm tries to be a good unix citizen. Use it with other commandline tools, like xsv for example:
atm worldcitiespop.csv filter "Population<20"|atm filter "Population>=10"|atm filter "Longitude<-50"|xsv select City,Population|xsv table
If you have rustup installed on your system you can simply run cargo install automat.
You can also run it via docker.
If you for example have a csv file called worldcitiespop.csv in your local directory:
docker run --rm -v $(PWD):/data oembot/automat ./atm /data/worldcitiespop.csv filter "Population<10"
The benchmarks have been created with hyperfine.
Now of course keep in mind that these benchmark are not all that useful, especially since they have been measured using time. Which is notoriously coupled to pretty much everything on the machine that is running those benchmarks. It really is just to show a very rough ballpark measurement.
| Command | Mean [s] | Min [s] | Max [s] | Relative |
|---|---|---|---|---|
target/release/atm test.csv filter "Population<1000000" |
1.352 ± 0.009 | 1.341 | 1.366 | 1.00 |
| Command | Mean [s] | Min [s] | Max [s] | Relative |
|---|---|---|---|---|
target/release/atm test.csv filter "Population<1000000" |
1.360 ± 0.010 | 1.346 | 1.379 | 1.00 |