FAIR Constrained Spectral Clustering Analysis
Budget: ₹600 – ₹1,500 INR
I am seeking a skilled data scientist to perform FAIR Constrained Spectral Clustering on several UCI ML datasets, including IRIS, HEPATITIS, CRIME, and WINE. The task involves running 100 iterations for each dataset, averaging the results, and varying the number of clusters (k) from 0 to 15. Proper documentation will be provided for the algorithms used.
After completing the FAIR Constrained Spectral Clustering, you will also need to perform unnormalized SC, normalized SC, fair normalized SC, and constrained SC (without fairness) on the same datasets to enable a comprehensive comparison of results(balance, rand index, no. of satisfied constraints, cost).
Note: YOU HAVE TO USE GOOGLE COLLAB FOR THIS AND SHARE THE WHOLE CODE LINK WITH ACCESS GIVEN.
Key Requirements:
- Conduct 100 iterations of FAIR Constrained Spectral Clustering and average the results.
- Test with k values ranging from 0 to 15.
- Perform additional clustering techniques for comparison: unnormalized SC, normalized SC, fair normalized SC, and constrained SC (without fairness).
- Utilize provided documentation for algorithm implementation.
Ideal Skills and Experience:
- Proficiency in Python, R, or Matlab for data analysis and clustering.
- Strong understanding of spectral clustering techniques and fairness metrics.
- Experience with UCI ML datasets and clustering evaluation metrics.
- Ability to measure effectiveness using rand index (not adjusted), balance, number of satisfied constraints, and cost.
The primary goal is to achieve the best results for FAIR Constrained Spectral Clustering compared to other methods, with a focus on balance and rand index (not adjusted) as the most important fairness metrics.
After completing the FAIR Constrained Spectral Clustering, you will also need to perform unnormalized SC, normalized SC, fair normalized SC, and constrained SC (without fairness) on the same datasets to enable a comprehensive comparison of results(balance, rand index, no. of satisfied constraints, cost).
Note: YOU HAVE TO USE GOOGLE COLLAB FOR THIS AND SHARE THE WHOLE CODE LINK WITH ACCESS GIVEN.
Key Requirements:
- Conduct 100 iterations of FAIR Constrained Spectral Clustering and average the results.
- Test with k values ranging from 0 to 15.
- Perform additional clustering techniques for comparison: unnormalized SC, normalized SC, fair normalized SC, and constrained SC (without fairness).
- Utilize provided documentation for algorithm implementation.
Ideal Skills and Experience:
- Proficiency in Python, R, or Matlab for data analysis and clustering.
- Strong understanding of spectral clustering techniques and fairness metrics.
- Experience with UCI ML datasets and clustering evaluation metrics.
- Ability to measure effectiveness using rand index (not adjusted), balance, number of satisfied constraints, and cost.
The primary goal is to achieve the best results for FAIR Constrained Spectral Clustering compared to other methods, with a focus on balance and rand index (not adjusted) as the most important fairness metrics.