Model evaluation dashboard for classification metrics

What this project does

This project constructs a small, reproducible evaluation pipeline for multiclass classification and produces a read-only evaluation view suitable for a dashboard. It shows per-class precision, recall and F1, confusion matrices, score curves where feasible, and compact exportable error summaries. The code also includes a tiny, hand-calculated fixture to verify behaviour when some classes receive zero predicted positives.

Data

The project uses the built-in sklearn iris dataset (3 classes, four numeric features). This is a small, well-known dataset used here only as a deterministic example; no external data is downloaded.

How to build it

1. Create a Python 3.11+ virtual environment.
2. Install dependencies: pip install -r requirements.txt
3. Open project.py to inspect the pipeline (train/test split, LogisticRegression classifier, metrics computation).
4. Run the script to produce a JSON evaluation object and a compact error summary.

How to run

Run the module directly with Python. It prints a JSON object produced by run(). Example:

python project.py

The run function accepts run(seed: int = 42) -> dict so you can change the random_state.

How to test

Run pytest in the project directory. The test suite imports run and checks keys, ranges, determinism and a conservative accuracy floor.

Results

Measured by running python project.py (seed 42) on 2026-10-23; the tests in test_project.py assert the floor.

Metric	Value
dataset_version	sklearn_iris_v1
accuracy	0.9333
macro_f1	0.9333
roc_auc_ovr	0.9956

Also returned (see python project.py output): per_class (object), confusion_matrix (3 items), exportable_error_summary (3 items).

Read the roc_auc_ovr value with care: the data here is small and synthetic with a planted, learnable structure, so a near-perfect score shows the pipeline works, not that the method would reach this on real data. See Limitations.

Limitations

This is a small, demonstration pipeline using a built-in dataset. The output is a read-only JSON-style view intended for further integration into a dashboard; it does not include a web UI. ROC/AUC computations are provided in a one-vs-rest form for the multiclass problem and require probability outputs; some score curves are less interpretable for small datasets. The hand-calculation fixture is intentionally tiny to exercise edge cases (absent predicted classes) and is not a substitute for extensive external validation.

Next steps

• Add a simple front-end (static HTML/JS) to render the returned JSON as interactive tables and plots.
• Provide a loader for a CSV import format matching the example schema (ids, true_label, pred_label, pred_score).
• Extend evaluation to grouped or subpopulation metrics (by group id) and save exportable CSV summaries for downstream review.

---

Full case study, build steps and evaluation checks: Model evaluation dashboard for classification metrics on networkershome.com

Built by Networkers Home · Learn the skills in the AI Engineer Course · Part of 100 AI projects with code