Skip to content

Regression with support vector machines (SVM). Clustering with DBSCAN. Identifies factors contributing to mental health help seeking behaviours in the tech industry

License

Notifications You must be signed in to change notification settings

ShrutiAppiah/MLforMentalHealth

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

24 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Paper & associated research

Read the associated IEEE-style research paper

Goal

To gauge which employer-related factors contribute to employees in the tech industry are with seeking help related to mental health issues.

Execution strategy

We used PCA to reduce the dimensionality of survey responses in the dataset, then clustered these responses with DBSCAN. To classify, we compared two different methods - a Naives Bayes Classifier and a Support Vector Machine (SVM).

The SVM ultimately performed better. SVMs are a perfect candidate given the relatively small volume of our dataset.

  • Identify clusters against the mental illness binary (i.e. the 'Do you currently have a mental health disorder?') while considering all other attributes at once
  • Identify contrasting clusters (YES/NO/No Preference) based on the mental illness binary while considering each attribute one by one iteratively.

    Use DBSCAN clustering to identify and separate data into different clusters. Obtain the number of clusters. To classify data, run SVM.

Results

Naive Bayes Accuracy

SVM Accuracy

Learnings

  • The dataset used was fairly limiting. Our testing and training models didn't perform optimally with both Navie Bayes nor SVM classifiers as a result.
  • We could've invested more effort into developing more effective de-noising strategies for the dataset.
  • There needs to be an extra step of data clean-up prior to PCA dimensionality reduction. This helps eliminate extreme outliers.

Dataset

https://www.kaggle.com/osmi/mental-health-in-tech-survey
https://www.kaggle.com/osmi/mental-health-in-tech-2016/version/1

Contributors

Shruti Appiah, Sam Barnard, and Jonathan Deiven

License

License: MIT

Copyright © 2017 Shruti Appiah

About

Regression with support vector machines (SVM). Clustering with DBSCAN. Identifies factors contributing to mental health help seeking behaviours in the tech industry

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published