This is an attempt to predict diseases from the given symptoms. A decision tree was trained on two datasets, one had the scraped data from here.
This dataset is uncleaned so preprocessing is done and then model is trained and tested on it.
Next another decision tree was also trained on manually created dataset which contains both training and testing sets. This data is cleaned and extensive and hence learning was more accurate. The exported decision tree looks like the following :
Head over to Data-Analyis.ipynb to follow the whole process.
- Create a web service in Flask for disease predictions using the trained model.
- Perform Affinity Analyis to observe which symptoms usually occur together.