README.md

Thyroid Disease Dataset

Please refer to original source for more details.

Dataset Description

The Thyroid Disease dataset comprises medical records related to thyroid conditions. Supplied by the Garavan Institute and J. Ross Quinlan from the New South Wales Institute, Sydney, Australia in 1987, this dataset is widely used for diagnosing thyroid disorders. It contains 3,772 instances with 30 attributes, including both continuous and discrete features.

Citation

Please cite the dataset as follows:

Thyroid disease records supplied by the Garavan Institute and J. Ross Quinlan, New South Wales Institute, Sydney, Australia. 1987.

Dataset Structure

• Number of Instances: 3,772
• Number of Attributes: 30
• Attribute Types:
  • Continuous: 7 (1 integer, 6 real)
  • Discrete: 23
• Missing Values: 6,064 (5.4% of the dataset)

Features

The dataset includes the following features:

Feature	Type	Description	Possible Values
age	Integer	Age of the patient	Continuous values
sex	Categorical	Gender of the patient	M, F
on thyroxine	Categorical	Whether the patient is on thyroxine medication	f (false), t (true)
query on thyroxine	Categorical	Whether there is a query on thyroxine usage	f, t
on antithyroid medication	Categorical	Whether the patient is on antithyroid medication	f, t
sick	Categorical	Whether the patient is sick	f, t
pregnant	Categorical	Whether the patient is pregnant	f, t
thyroid surgery	Categorical	Whether the patient has undergone thyroid surgery	f, t
I131 treatment	Categorical	Whether the patient has received I131 treatment	f, t
query hypothyroid	Categorical	Query on hypothyroid status	f, t
query hyperthyroid	Categorical	Query on hyperthyroid status	f, t
lithium	Categorical	Whether the patient is taking lithium	f, t
goitre	Categorical	Presence of goitre	f, t
tumor	Categorical	Presence of a tumor	f, t
hypopituitary	Categorical	Hypopituitary condition status	f, t
psych	Categorical	Psychological condition status	f, t
TSH measured	Categorical	Whether TSH was measured	f, t
TSH	Real	Thyroid Stimulating Hormone level	Continuous values
T3 measured	Categorical	Whether T3 was measured	f, t
T3	Real	Triiodothyronine level	Continuous values
TT4 measured	Categorical	Whether TT4 was measured	f, t
TT4	Real	Total T4 level	Continuous values
T4U measured	Categorical	Whether T4U was measured	f, t
T4U	Real	T4 Uptake	Continuous values
FTI measured	Categorical	Whether FTI was measured	f, t
FTI	Real	Free Thyroxine Index	Continuous values
TBG measured	Categorical	Whether TBG was measured	f, t
TBG	Real	Thyroxine Binding Globulin	Continuous values
referral source	Categorical	Source of referral	WEST, STMW, SVHC, SVI, SVHD, other
Class	Categorical	Diagnosis class	sick, negative

Data Statistics

• Missing Values: 6,064 out of 113,160 total data points (5.4%)
• Distinct Values per Feature:
  • age: 93 distinct values (2% missing)
  • sex: 2 distinct values (4% missing)
  • on thyroxine: 2 distinct values (0% missing)
  • query on thyroxine: 2 distinct values (0% missing)
  • on antithyroid medication: 2 distinct values (0% missing)
  • sick: 2 distinct values (0% missing)
  • pregnant: 2 distinct values (0% missing)
  • thyroid surgery: 2 distinct values (0% missing)
  • I131 treatment: 2 distinct values (0% missing)
  • query hypothyroid: 2 distinct values (0% missing)
  • query hyperthyroid: 2 distinct values (0% missing)
  • lithium: 2 distinct values (0% missing)
  • goitre: 2 distinct values (0% missing)
  • tumor: 2 distinct values (0% missing)
  • hypopituitary: 2 distinct values (0% missing)
  • psych: 2 distinct values (0% missing)
  • TSH measured: 2 distinct values (0% missing)
  • TSH: 287 distinct values (10% missing)
  • T3 measured: 2 distinct values (0% missing)
  • T3: 69 distinct values (2% missing)
  • TT4 measured: 2 distinct values (0% missing)
  • TT4: 241 distinct values (6% missing)
  • T4U measured: 2 distinct values (0% missing)
  • T4U: 146 distinct values (4% missing)
  • FTI measured: 2 distinct values (0% missing)
  • FTI: 234 distinct values (6% missing)
  • TBG measured: 1 distinct value (0% missing)
  • TBG: 3,772 distinct values (0% missing)
  • referral source: 5 distinct values (0% missing)
  • Class: 2 distinct values (0% missing)

Usage

This dataset is ideal for training classification models to diagnose thyroid disorders based on patient medical records. It can be used for tasks such as binary classification, feature analysis, and missing data handling techniques.