24679-hw2-shoe-owner-autogluon

ypolatog/24679-hw2-shoe-owner-autogluon

24679-hw2-shoe-owner-autogluon is a tabular classification model from ypolatog. Use it for the tabular classification task on the model card, and read the license before you ship it in a product. The card lists the license as other.

Predicts which of three roommates owns a shoe from its length, shaft height, type, colour and closure. Trained with AutoGluon TabularPredictor on ArinRoths/shoe-data, a classmate's HW1 dataset.

Downloads · 30 days

Access

Public

Updated Sep 21, 2026

Repo size

6.9 MB

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.zip6.9 MB · 100%

At a glance

Task: Tabular Classification
License: other
Access: Public
Created: Sep 21, 2026
Updated: Sep 21, 2026
SHA: e4bb4752

Model card

From the Hugging Face model README

Shoe owner classifier — 24-679 HW2 Part 1

Predicts which of three roommates owns a shoe from its length, shaft height, type, colour and closure. Trained with AutoGluon TabularPredictor on ArinRoths/shoe-data, a classmate's HW1 dataset.

Results

Primary evidence is grouped 5-fold cross-validation over all 25 original shoes; folds are split by parent_id so augmented variants never cross a fold boundary.

Metric	Cross-validated	Held-out test (n=5)
Balanced accuracy	0.588 +/- 0.217	0.167
Accuracy	0.600	0.200
Macro F1	0.564	0.167

Chance is 0.333. Majority baseline on test: 0.200 (always predicts Arin).

Selected model

NeuralNetTorch_BAG_L1/74058_00000 (StackerEnsembleModel), chosen by a 20-trial random search over GBM, CAT, XGB, NN_TORCH within a 900 s budget (756 s used).

Hyperparameter	Value
`activation`	`relu`
`dropout_prob`	`0.1`
`embed_exponent`	`0.56`
`embedding_size_factor`	`1.0`
`epochs_wo_improve`	`None`
`hidden_size`	`32`
`learning_rate`	`0.0003`
`loss_function`	`auto`
`max_batch_size`	`512`
`max_embedding_dim`	`100`
`num_epochs`	`50`
`num_layers`	`2`
`optimizer`	`adam`
`proc.embed_min_categories`	`4`
`proc.impute_strategy`	`median`
`proc.max_category_levels`	`100`
`proc.skew_threshold`	`0.99`
`seed_value`	`0`
`use_batchnorm`	`False`
`use_ngram_features`	`False`
`weight_decay`	`1e-06`
`y_range`	`None`
`y_range_extend`	`0.05`

Intended use and limitations

Coursework for CMU 24-679 Design AI; not for any real decision. The dataset has 25 original shoes from one household, so the model does not generalise beyond these three people's shoe collections, and the test split is 5 shoes — far too few to judge on alone, which is why the cross-validated figure is quoted first.

The source dataset declares no license. Reuse should be cleared with its author, ArinRoths.

Reproducing

Seed 24679, AutoGluon 1.6.1. metrics.json in this repository holds every number above. Load with TabularPredictor.load() after unzipping autogluon_predictor_dir.zip.

Task: Tabular Classification
License: other
Created: Sep 21, 2026
Updated: Sep 21, 2026