Dark Mode
Currency Forgery Detection Data
Finance & Banking Analytics
Tags and Keywords
Trusted By
"No reviews yet"
Free
About
Data pertaining to banknote authentication is provided, derived from images of both genuine and forged specimens. Features were extracted from 400x 400 pixel gray-scale images (captured at approximately 660 dpi) using the Wavelet Transform tool. This data product is designed to facilitate the development and testing of robust classification algorithms capable of distinguishing between authentic currency and counterfeits, making it highly valuable for anti-fraud modelling.
Columns
The data product contains five columns detailing continuous features derived from the banknote image analysis and a final integer class label:
- variance: A continuous value representing the variance observed in the Wavelet Transformed image.
- skewness: A continuous value representing the skewness observed in the Wavelet Transformed image.
- curtosis: A continuous value representing the curtosis observed in the Wavelet Transformed image.
- entropy: A continuous value representing the entropy of the original image.
- class: An integer label indicating the status of the specimen (e.g., 0 for genuine, 1 for forged).
Distribution
The dataset is structured as a single file, BankNoteAuthentication.csv, with a size of approximately 46.44 kB. It contains 5 columns and 1372 valid records. The dataset is notable for its quality, as zero missing or mismatched values are reported. The class labels are distributed with 762 counts for one class (0.00 - 0.10) and 610 counts for the alternative class (0.90 - 1.00).
Usage
Ideal applications for this data product include training supervised machine learning models for binary classification. It is widely used for benchmarking different classification algorithms focused on fraud detection, especially in banking and finance sectors. The unique feature extraction methodology (Wavelet Transform) also makes it suitable for advanced digital image processing studies.
Coverage
The underlying data relates to banknote-like specimens. This dataset was received in August 2012. Given its origin related to academic research on print inspection, it focuses purely on the statistical features derived from the imagery. No further updates are expected for this dataset.
License
CC0: Public Domain
Who Can Use It
This dataset is essential for machine learning practitioners and data scientists building classification models. It is highly useful for researchers and students in universities and colleges studying computer vision, feature engineering, and automated fraud detection systems. Experts focusing on banking security and currency authenticity systems would also find it valuable.
Dataset Name Suggestions
- Wavelet-Based Banknote Features
- Currency Forgery Detection Data
- UCI Banknote Authenticator
- Image Feature Classification Benchmark
Attributes
Original Data Source: Currency Forgery Detection Data
QUALITYLoading...