expview 0.1.1

Experimental data manager and visualization tools

ExpView

⚠️ Currently under active development.

Overview

ExpView is a terminal-based analytics and visualization tool inspired by Excel.
Designed for developers and data scientists, it displays pandas DataFrames directly in the terminal using the Curses library — with fast navigation, shortcut keys, and instant refresh.

ExpView can integrate with experiment pipelines: pass variable configurations to external programs, collect CSV/TSV/JSON outputs, manage configurations, and visualize results with matplotlib to analyze parameter effects and performance trends.

---

Installation

pip install expview

This installs ExpView and its dependencies.

---

Quick Start

To open a dataset, run:

expview data.csv

If you have multiple datasets in a folder, just run:

expview

---

Using ExpView to Generate Experiment Results

Example usage:

# my_experiment.py
import random
from expview import experiment, cli

@experiment
def my_experiment(run_args, exp_vars, results):
    """Example experiment function."""
    print(f"--- Experiment {run_args['expid']} ---")
    for k, v in exp_vars.items():
        print(f"{k}: {v}")
    results["accuracy"] = round(random.uniform(0.4, 0.99), 2)
    print("accuracy:", results["accuracy"])

if __name__ == "__main__":
    cli()

Then call your program like this:

python my_experiment.py run --var1=val1/val2 --var2=val1/val2

This runs the experiment with all combinations of input variables and creates four test results in logs directory. For full documentation, refer to the GitHub homepage.

---

Exploring Results

Run expview inside the experiment or logs directory or any subdirectory containing the result files.
ExpView merges the files and displays results in a navigable terminal table:

expview

---

Plotting

ExpView supports interactive plotting. Example workflow:

1. Select a column as x-axis (Shift + X)
2. Select another column as y-axis (Shift + Y)
3. Optionally, select a column for legend (Shift + T)
4. Open the command prompt (:) and type line to generate a plot.

For full documentation, visit the GitHub repository.

---

License

MIT License © 2025 Ahmad Pouramini