This project is about predicting covid positive diagnosis by variables such as symptoms. Data from the Israeli Ministry of Health is used to build the classifier. The objective of the project is to predict patients which positive results and analyze the relationship between symptoms and positive cases.
Things I Like:
The project included a solid discussion of fairness and weapon of math destruction. The section included a chi-sq test and other suggestions to prevent the model from being harmful.
The project includes a variety of relevant models with explanations of the reason behind using the model. Mentioning the false-negative rate and false-positive rate also helped with the analysis.
Considering the case of covid and influenza added more depth to the project. It was also great to see the reasoning behind including the section and explaining the importance of the distinction.
Area for Improvement
Although I understand that finding data related to Covid and Influence may be difficult, I believe it would have been better if the project consisted of a dataset that compared two similar samples. People from Israel and West Virginia probably have different living styles, so it would have made more sense to create models from similar samples.
For the objective of this project, I didn't really see the need for figure 1. The way someone gets infected by COVID today is most likely not going to change from the way someone gets infected by COVID a year later.
When fixing the imbalanced dataset further explanation about the potential bias and how it can affect generalization would be useful.
This project is about predicting covid positive diagnosis by variables such as symptoms. Data from the Israeli Ministry of Health is used to build the classifier. The objective of the project is to predict patients which positive results and analyze the relationship between symptoms and positive cases.
Things I Like:
Area for Improvement