Skip to content

Getting Started

Prerequisites

  • Python 3.12+
  • R (required for Propensity Score Matching via MatchIt)
  • Git

Installation

git clone https://github.com/CTN-0094/Pipeline.git
cd Pipeline
pip install -r requirements.txt

import os
if not os.path.exists("/content/Pipeline"):
    !git clone https://github.com/CTN-0094/Pipeline.git
    os.chdir("/content/Pipeline")

!pip install -r requirements.txt

Open In Colab

Running the Pipeline

Basic Usage

python3 run_pipelineV2.py --data <path/to/data.csv> --outcome <outcome_name> -d ./results

Examples

python3 run_pipelineV2.py \
  --data data.csv \
  --outcome ctn0094_relapse_event \
  -d ./results

# Loop through seeds 5–10 across all outcomes
python3 run_pipelineV2.py \
  --data data.csv \
  -l 5 10 \
  -d ./results

python3 run_pipelineV2.py \
  --data data.csv \
  -l 5 10 \
  -o ctn0094_relapse_event Ab_ling_1998 \
  -d ./results

# Run preprocessing + PSM only, skip model training
python3 run_pipelineV2.py \
  --data data.csv \
  --outcome ctn0094_relapse_event \
  --data_only \
  -d ./results

CLI Reference

Argument Default Description
--data (required) Path to cleaned input dataset (CSV)
-o / --outcome all Outcome(s) to run
-l / --loop — Min and max seed for multi-seed runs
-d / --dir "" Output directory for results
--type — Endpoint type for custom outcomes: logical, integer, survival
--majority 1 Value representing the majority group in the PSM column
--split RaceEth Column used to define majority/minority groups
--match age is_female Columns to match on during PSM
--group_size 500 Size of each PSM group
--heldout_size 100 Size of the held-out evaluation set
--heldout_set_percent_majority 58 Percent majority in the held-out set
--data_only False Skip model training, save preprocessed data only
-p / --prof — Profiling mode: simple or complex

Viewing Results

After a run, your output directory contains:

results/
├── logs/                     # Execution logs per run
├── heldout_predictions/      # Held-out set predictions (CSV)
├── subset_predictions/       # Subset predictions by ratio (CSV)
├── heldout_evaluations/      # Metrics on the held-out set (CSV)
├── subset_evaluations/       # Metrics per demographic ratio (CSV)
└── experiments/              # Experiment metadata (JSON)

Reading the evaluation CSVs

Each row in an evaluation CSV corresponds to a different majority/minority demographic ratio. The rightmost columns are the evaluation metrics. Compare rows to see how model performance changes as the training cohort composition shifts.

R Package Installation

When you run PSM for the first time, you will be prompted to install R packages:

Would you like to install MatchIt?
1: Yes
2: No

Selection: 1

Select 1 for both MatchIt and optmatch. This only happens once.