Dark Mode
Forensic Glass Composition Dataset
Data Science and Analytics
Tags and Keywords
Trusted By
"No reviews yet"
Free
About
This dataset focuses on the identification of different types of glass, primarily for forensic science applications. Originating from the USA Forensic Science Service, it was created by B. German and donated by Vina Spiehler in September 1987. The data was gathered to assist in criminological investigations, particularly in correctly identifying glass fragments found at crime scenes, which can serve as crucial evidence. The dataset has been used in comparative tests of classification algorithms, including rule-based systems like BEAGLE, nearest-neighbour algorithms, and discriminant analysis, where the nearest-neighbour method performed well relative to the rule-based system.
Columns
The dataset comprises 10 continuously valued attributes, including an ID number, plus a class attribute indicating the type of glass. There are no missing attribute values.
- Id: A unique identification number for each instance, ranging from 1 to 214.
- RI: Refractive index, a measure of how light propagates through the glass.
- Na: Sodium content, measured as weight percent in the corresponding oxide.
- Mg: Magnesium content, measured as weight percent in the corresponding oxide.
- Al: Aluminium content, measured as weight percent in the corresponding oxide.
- Si: Silicon content, measured as weight percent in the corresponding oxide.
- K: Potassium content, measured as weight percent in the corresponding oxide.
- Ca: Calcium content, measured as weight percent in the corresponding oxide.
- Ba: Barium content, measured as weight percent in the corresponding oxide.
- Fe: Iron content, measured as weight percent in the corresponding oxide.
- Type of glass: The classification attribute, indicating the glass type:
- 1: building windows (float processed)
- 2: building windows (non-float processed)
- 3: vehicle windows (float processed)
- 4: vehicle windows (non-float processed) – none in this dataset
- 5: containers
- 6: tableware
- 7: headlamps
Distribution
This tabular dataset consists of 214 instances (rows) and 11 attributes, including the class attribute, all of which are continuously valued. The dataset is provided in a CSV format, specifically
glass.csv, and is approximately 10.6 kB in size. The class distribution across the 214 instances is as follows:- Window glass (163 instances):
- Float processed: 87 instances (70 building windows, 17 vehicle windows)
- Non-float processed: 76 instances (76 building windows, 0 vehicle windows)
- Non-window glass (51 instances):
- Containers: 13 instances
- Tableware: 9 instances
- Headlamps: 29 instances
Usage
This dataset is ideally suited for applications in machine learning, particularly for classification tasks and rule induction. It is highly valuable for:
- Forensic Science Research: Developing and testing methods for glass identification in criminological investigations.
- Algorithm Comparison: Benchmarking and comparing the performance of various classification algorithms, such as nearest-neighbour, rule-based systems, and discriminant analysis.
- Material Science Studies: Analysing the chemical composition of different glass types and their classification.
- Educational Purposes: Providing a practical example for teaching data classification, attribute analysis, and the application of machine learning in real-world scenarios.
Coverage
The data originates from the USA Forensic Science Service. The collection date is September 1987. The scope of the dataset covers various types of glass commonly found in building windows, vehicle windows, containers, tableware, and headlamps. It is noted that there are no instances of 'vehicle windows (non-float processed)' within this particular database.
License
CC0: Public Domain
Who Can Use It
This dataset is beneficial for a range of users, including:
- Forensic scientists and researchers: For developing and refining techniques in glass evidence analysis.
- Data scientists and machine learning engineers: For building and evaluating classification models.
- Academics and students: For educational purposes in fields such as data science, chemistry, and criminology.
- Anyone interested in material classification: To explore the relationship between chemical composition and glass type.
Dataset Name Suggestions
- Forensic Glass Composition Dataset
- Glass Type Classification Data
- Chemical Glass Identification Data
- Criminological Glass Analysis Dataset
- Material Property Glass Classification
Attributes
Original Data Source: Forensic Glass Composition Dataset
QUALITY