Name: hiv-retention-public
Owner: Data Science for Social Good
Description: null
Created: 2018-02-12 00:53:17.0
Updated: 2018-02-13 17:15:54.0
Pushed: 2018-02-13 17:27:22.0
Homepage: null
Size: 33
Language: Python
GitHub Committers
| User | Most Recent Commit | # Commits |
|---|
Other Committers
| User | Most Recent Commit | # Commits |
|---|
Repo for HIV Retention DSaPP project (Public code).
To create the tables needed for modeling, execute:
on etl/cli.py etl/queries/clean_data.sql etl/queries/create_states_table.sql etl/queries/expert_and_demographic_features.sql
Run all the code in DEV_load_cdph_common_schema.ipynb
cli.py accesses etl/inventory.yaml to find out which CSVs have to get used for the tables in raw. etl/queries/clean_data.sql cleans the raw tables and moves them to staging. etl/queries/create_states_table.sql creates the states table that is used by Triage to know when an individual should be included in the modeling process. etl/queries/expert_and_demographic_features.sql creates two feature tables.
In order to create the events table, execute:
on etl/label_maker.py
The events table gets used by Triage to create the labels. Once the tables are created, you can run a Triage experiment:
line/Triage_Run.py pipeline/parameter_config.yaml
The experiment_config tells Triage which features and models to use. When you pass the true argument, the experiment adds the regex features.
Triage stores its results in the DB. Some common queries for the analysis of model performance can be found in queries/comp.sql and queries/performance_eval.sql.