zacCMU/2025-24679-pens-tabular-dataset

Pens Dataset

This is a simple, manually-collected tabular dataset containing the physical attributes of 33 unique writing pens. It is designed for straightforward binary classification tasks. The features include quantitative measurements like length, and qualitative features like ink color, body material, and brand.

The primary target variable is cap_presence, a binary indicator (1 or 0) signifying whether a pen has a separate cap or not (e.g., a capped pen vs. a retractable click-pen).

An augmented configuration is also provided, containing 301 synthetic samples generated from the original data to facilitate model training on a larger dataset.

---

Supported Tasks

• tabular-classification: The dataset is ideal for predicting the cap_presence feature based on the other attributes. This can be used as a beginner-friendly project for training models like Logistic Regression, Decision Trees, or simple Neural Networks.

---

Languages

The data is in English (en).

---

dataset-structure Structure

Data Instances

A typical row in the original split looks like this:

{
  "Unnamed: 0": 0, (index)
  "length_cm": 14.5,
  "ink_color": "Black",
  "body_material": "Plastic",
  "cap_presence": 1,
  "brand": "BIC"
}