Scribd
Upload
5 views3 pages

? Basic Questions

The project aims to predict patient risk levels (High Risk or Low Risk) using basic health data through a binary classification problem with a Support Vector Classifier (SVC) and RBF kernel. It utilizes features such as heart rate and oxygen level from a dataset of around 7000 records, with preprocessing steps including handling missing values and feature engineering. The system provides predictions with confidence scores and visualizations, while acknowledging limitations like limited data and lack of clinical validation.

Uploaded by

debmitaah5
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
5 views3 pages

? Basic Questions

The project aims to predict patient risk levels (High Risk or Low Risk) using basic health data through a binary classification problem with a Support Vector Classifier (SVC) and RBF kernel. It utilizes features such as heart rate and oxygen level from a dataset of around 7000 records, with preprocessing steps including handling missing values and feature engineering. The system provides predictions with confidence scores and visualizations, while acknowledging limitations like limited data and lack of clinical validation.

Uploaded by

debmitaah5
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

🔹 Basic Questions

1. What is your project about?​


Predicts whether a patient is High Risk or Low Risk using basic health data.

2. What type of problem is this?​


It is a binary classification problem.

3. What algorithm did you use?​


Support Vector Classifier (SVC) with RBF kernel.

4. Why did you choose SVC?​


Because it works well with non-linear data and gives good accuracy.

🔹 Dataset & Features


5. What features are used?​
Heart rate, respiratory rate, temperature, oxygen level, age, gender.

6. What is the target variable?​


Risk category (High Risk / Low Risk).

7. How many records in dataset?​


Around 7000 records.

🔹 Preprocessing
8. How did you handle missing values?​
Used dropna() to remove them.

9. Why encoding is needed?​


ML models need numerical data.

10. What encoding techniques used?​


LabelEncoder (target), OneHotEncoder (gender).

11. Why scaling is important?​


To make all features equal in importance.
🔹 Feature Engineering
12. What is feature engineering?​
Creating new features from existing data.

13. Give example of new feature.​


Oxygen_Deficit = 100 - Oxygen.

14. Why feature engineering is important?​


Improves model performance.

🔹 Model & Training


15. What is RBF kernel?​
Helps model learn complex patterns.

16. What is class imbalance?​


One class has more data than the other.

17. How did you handle imbalance?​


Used class_weight = balanced.

🔹 Evaluation
18. What metrics used?​
Accuracy, Precision, Recall, F1 Score.

19. Why F1 score important?​


Balances precision and recall.

20. What is confusion matrix?​


Shows correct and incorrect predictions.

21. What is false negative?​


High-risk patient predicted as low-risk.

🔹 System & Output


22. What is your system capable of?​
Single & batch prediction with probability.
23. What tool used for UI?​
Streamlit.

24. What does your app show?​


Prediction, confidence score, and graphs.

🔹 Advanced / Important
25. What is hyperparameter tuning?​
Finding best model parameters.

26. Which method used for tuning?​


RandomizedSearchCV.

27. What is feature importance?​


Shows which feature affects prediction most.

28. Most important feature?​


Heart rate.

🔹 Limitations
29. What are limitations?​
Limited data, no clinical validation.

30. Future improvements?​


Use XGBoost, SMOTE, better explainability.

You might also like