From gnuplot to Matplotlib & Pandas - Bert Hubert's writings
a day ago
- The author transitions from gnuplot to Matplotlib and Pandas for data plotting and analysis, citing gnuplot's limitations outside its core strengths.
- Pandas and Matplotlib offer integrated data labeling and manipulation, contrasting gnuplot's need for external preprocessing with awk and shell scripts.
- The article demonstrates practical examples: loading vmstat data into Pandas, plotting context switches vs. interrupts, and handling time windows with rolling statistics.
- Pandas supports advanced features like stacked graphs, derivatives using diff() or np.gradient(), and resampling to combine datasets or reduce noise.
- Data selection in Pandas is flexible using boolean indexing or query() methods, enabling precise filtering for analysis.
- Jupyter notebook provides an accessible interface for Matplotlib and Pandas, with setup instructions and a GitHub repository of sample data and code.