SKILL.md
Jupyter Notebook Guide
A skill for using Jupyter notebooks effectively in research contexts. Covers notebook organization, reproducibility best practices, collaboration workflows, and integration with research computing infrastructure.
Notebook Organization
Recommended Structure
Every research notebook should follow a consistent structure:
01_data_collection.ipynb # Data acquisition and initial storage
02_data_cleaning.ipynb # Preprocessing, validation, transformations
03_exploratory_analysis.ipynb # EDA, descriptive statistics, initial plots
04_modeling.ipynb # Model training, evaluation, selection
05_results_visualization.ipynb # Publication-quality figures
06_supplementary.ipynb # Additional analyses, robustness checks
Cell Organization Within a Notebook
# === CELL 1: Header and metadata ===
"""
# Analysis: Effect of Treatment on Outcome Variable
Author: [Name]
Date: 2026-03-09
Data: experiment_results_v2.csv
Dependencies: pandas>=2.0, scipy>=1.11, matplotlib>=3.8
"""
# === CELL 2: Imports and configuration ===
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from scipy import stats
# Reproducibility
np.random.seed(42)
pd.set_option('display.max_columns', 50)
plt.rcParams.update({
'figure.figsize': (10, 6),
'figure.dpi': 150,
'font.size': 12,
'axes.titlesize': 14,
'savefig.dpi': 300,
'savefig.bbox': 'tight'
})
# === CELL 3: Data loading ===
DATA_PATH = '../data/raw/experiment_results_v2.csv'
df = pd.read_csv(DATA_PATH)
print(f"Loaded {len(df)} rows, {len(df.columns)} columns")
df.head()
Reproducibility Best Practices
Environment Management
Always pin your dependencies:
# Create environment from scratch
conda create -n research python=3.11
conda activate research
# Install and pin
pip install pandas==2.1.4 scipy==1.11.4 matplotlib==3.8.2 jupyterlab==4.0.9
# Export for reproducibility
pip freeze > requirements.txt
# Or use conda
conda env export --no-builds > environment.yml