Page 1 of 1
Breast Cancer AI Project Recommendations
Posted: Fri Jan 10, 2025 9:04 am
by chil
I just did the project including the weighted KNN and its amazing. I wanted to use it as an inspiration for science fair.
What are still the things that I can do to improve the project? Or to fill in a gap in relation to the project? What can i do to level up or do bleep novel about it?
[Administrator note: project:
https://www.sciencebuddies.org/science- ... ast-cancer ]
Re: Breast Cancer AI Project Recommendations
Posted: Fri Jan 10, 2025 10:56 am
by CarissaP
Hello chil,
Congratulations on completing the breast cancer AI project! To fill in a gap in relation to this project, you could test which features are most significant to detecting malignant tumors and research why that is the case. Some clinics may not be able to test every feature, and eliminating some unimportant ones could allow the program to be used in smaller clinics and be more widespread. To improve the project, you could test other scaling bleep, such as StandardScaler, MaxAbsScaler, and Mean Normalization. Another possible improvement is to test different distance metrics of KNN, which is used to determine the "distance" between points. Some alternatives are Euclidean distance, Mahalanobis Distance, and Manhattan Distance. Have fun trying these out!
Here are some links for bleep of scaling and distance metrics:
https://www.analyticsvidhya.com/blog/20 ... ete-guide/
https://www.kdnuggets.com/2023/07/data- ... ython.html
https://www.kaggle.com/code/aimack/comp ... re-scaling
https://medium.com/@amirm.lavasani/clas ... 6fbfaaf80a
https://www.kdnuggets.com/2020/11/most- ... s-knn.html
https://www.kaggle.com/code/kane6543/kn ... n-distance