stevenxxiu/sml_project_2

Statistical Machine Learning Project 2

sml_project_2

Statistical Machine Learning Project 2 For Part A, run the file R/part_a.r, changing the variable 'group' in line 112 with group_1, group_2, group_3 or group_4 for each plot. For Part B, run the file R/disssimilarities.r, R/finnCorrelation.r and predict_linreg.r.

sml_project_2's Issues

Investigate alternative similarity measures

Fatima, you had this paper on alternative similarity measures to the usual euclidean distance, perhaps you could try to implement this.

Ideas for part B

Example 1

Clustering migrant information to see which surburbs migrants are coming to. Also find out about education, do they produce the same clusters?

Professional and migrants.

Migrants and land use.

Population aged 70-80 near to hospital and nursing homes.

Population in school and secondary school.

Example 2

I'll do logistic regression with normalized features, or svm with feature selection, and figure out which features contribute to the result.

Some other ideas besides the examples:

Associations:

• Migration and education.
• Health and hospitals.

Try to find associations or similarities in clusters between the number of presentations to hospitals (there are like 3 features) and some other feature or set of features like profession, age, land use, etc. This way we could suggest that these presentations might be prevented by adding more education, or care centres, etc.

We can try to find how many disadvantaged people (single parents, poor people, people with disabilities, etc) are in an area and try to see which services lack there like care centres, hospitals, or maybe long traveling time to a hospital.

MDS in sklearn

Remove redundant features via correlation