Skip to content

building a model that determines whether or not the patient's symptoms indicate that the patient has hypothyroid.

Notifications You must be signed in to change notification settings

Mwadz/Hypothyroidism-Prediction

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 

Repository files navigation

Hypothyroidism Prediction

Objective

To build a model that determines whether or not the patient's symptoms indicate that the patient has hypothyroid.

Success metric

Built several models using SVM, Decision trees and Random Forest.

Context

Nairobi Hospital conducted a clinical camp to test for hypothyroidism. The data collected focused on Thyroid patients.A normal TSH and T4 is indicative of normal functioning of the thyroid gland, a low TSH and elevated T4 indicates hyperthyroidism, a low TSH and low T4 indicates secondary hypothyroidism, and a high TSH and low T4 indicates primary hypothyroidism.

Experimental Design

The process will entail:

i) Reading and exploring the given dataset.

ii) Defining the appropriateness of the available data to answer the given question.

iii) Finding and deciding what to do with outliers, anomalies, and missing data within the dataset.

iv) Performing exploratory data analysis while recording the observations.

v) Performing regression analysis. Incorporating categorical independent variables into the models(depending on whether necessary).

vi) Building the models in two parts.

vii) Performing evaluation of the models in the data

vii) Making a conclsion based on model analysis.

Data Relevance

The appropriate dataset to use for this project is that which contains data with information on patients from a clinical camp held by the Nairobi Hospital. The data provided is useful and very relevant.

About

building a model that determines whether or not the patient's symptoms indicate that the patient has hypothyroid.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published