Asset Tags
Projit allows you to tag your various assets and then use them when querying and displaying summaries of a project
Command Line Tagging
Add tags to any project asset with a simple commnand line option
>projit tag dataset train type=tabular,source=Kaggle
This command will create two tags for for the dataset train one for type with the value tabular and another for source with the value Kaggle
These tags will not be included by default when you list the datasets, but can be included by using an optional CLI switch, as follows:
>projit list datasets --tags source type
__Datasets________________________________________________________________________________
__Name____source___type______Path_________________________________________________________
train Kaggle tabular data/processed.csv
The command will now include the tags in the listing as is shown above.
Tagging Experiments
You can add tags to experiments using the CLI
>projit tag experiment "Initial Exp" algo=NaiveBayes
This command is adding an indication of the algorithm used in the named experiment. These tags could be for feature engineering, sampling or any ad hoc element of your data science methodology. The best practice is to determine a naming convention and Ontology for tagging your experiments so that you can make meaningful comaprisons across projects.
This tagging can also be done programmatically using the python API.
import projit as pit
project = pit.projit_load()
tags = {"algo":"NaiveBayes"}
project.add_experiment("Initial Exp", "experiments/exp_one.py", tags=tags)
Tagging can also be applied when starting the experiment.
import projit as pit
project = pit.projit_load()
tags = {"algo":"NaiveBayes"}
exec_id = project.start_experiment("Initial Exp", "experiments/exp_one.py", params={}, tags=tags)
#
# INSERT ALL EXPERIMENT CODE HERE
#
project.end_experiment("Initial Exp", exec_id, hyperparams={})
Tags can also be include when you list the experiments using the CLI:
>projit list experiments --tags algo
__Experiments_______________________________________________________________________________
__Name_____________algo________Runs__MeanTime____Path_______________________________________
Initial Exp NaiveBayes 4 42s experiments/exp_one.py
Second Exp ExtraTrees 2 19s experiments/exp_two.py
This will produce a table that includes the specified tags along with the default information about your experiments.