Science Buddies: "Ask an Expert"

I completed the AI detection of breast cancer and I wanted to make something different from it.

Can you suggest me any variations or recommendations for those who wants to further improve it aside from the stated variations on the project at the end of it?

Can I change the current csv file dataset with other kinds of dataset?

Hi - It's great that you did the Breast Cancer AI project (https://www.sciencebuddies.org/science- ... ast-cancer) and explored the KNN algorithm with it. Was this for a school project or science fair?

You already did the optional extension on weighted KNN?

If you want to consider expanding your exploration of AI, there are a number of other projects that use KNN or other machine learning algorithms. There are projects on tick identification, Titanic predictions, air quality, water quality, and more. You can view many of these projects in one place here: https://www.sciencebuddies.org/blog/art ... e-projects

The one on the Titanic also uses KNN.

I think there are two not yet listed on that page, so just in case these are of interest:

Air Quality: https://www.sciencebuddies.org/science- ... ir-quality

Thyroid Recurrence: https://www.sciencebuddies.org/science- ... oid_cancer

Amy
Science Buddies

Hello!

This is for school project.
I haven't tried weighted KNN yet. Thanks for that and your suggestions. Will check them out.

By the way, one last question:
Can I change the csv dataset file link to another dataset??

Great! Since this is for a school project, you probably want to stick with this same exploration right now. The weighted KNN is noted as the "variation," but if you scroll up from that, there is a longer explanation of what is involved. Look for this section: "Experimenting with Weighted KNN (Optional)"

I think that would be your best approach to doing "more" right now with this same project.

You could use another data set if it matched up in terms of the data included/called in the code. This project, as you know, uses a dataset from Dr. William H. Wolberg from the University of Wisconsin Hospitals in Madison -- the Wisconsin Diagnostic Breast Cancer (WDBC) dataset. It contains data about the nuclei of cells found in breast tumor samples. The project lists 10 features of the nuclei in the samples that are in the data.

You would likely have to do some additional preprocessing of your data and manipulation of the code in the notebook to work with another data file. The project itself is written for the specific data set noted.

Would you be interested in sharing a bit more about your project after it's all done?

Amy
Science Buddies

Hello thank you so much for your fast response! Yes I will update you about the results of this science project this weekend and would try to consider the weighted KNN and different kind of datasets.

How about if I try to deploy the model into a web app? Like I can upload images of breast tumor samples there and it will predict if it is benign or malignant?

Replying for the updates.

The project itself isn't set up to deploy to a web app, but there are ways to publish colab notebooks as web apps. We don't have guidance on this (at this point) as the projects really are an introduction to different algorithms and aspects of working with machine learning.

Amy
Science Buddies