idea or code and explanation in matlab for classification for a disease dataset
I think the most useful framework for this purpose is MeSH - you can also try ICD.
If you are trying to possibly handle this as a typical Machine Learning based classification problem, I would suggest the MATLAB's Classification Learner Toolbox, which provides an easy way to simply load the variables (features) from your workspace (which would be your features), and use it to predict the target (which in your case possibly might be a binary 0/1 for normal/disease segmentation, or if you are classifying multiple diseases, then it would be a multi-class classification problem in that sense).
You can even export the models afterwards to the workspace, save it, generation the MATLAB code for the trained model and much more. This is possibly the easiest way to handle classification tasks in MATLAB at present in my opinion, with minimal coding.
R or Python may be also appropriate for simple topics on disease datasets. Many examples can be found on Kaggle datasets.
Hi, Suresh,
If you are planning a CART model then--after you get the model--the following (manual) procedure is used to prune the model to obtain maximum accuracy:
- https://odajournal.com/2019/04/14/optimizing-suboptimal-classification-trees-matlab-cart-model-predicting-probability-of-lower-limb-prosthesis-users-functional-potential
This is true for all tree-based classification models:
- https://odajournal.com/2019/01/22/maximizing-classification-accuracy-of-cart-recursive-partitioning-tree-models-using-optimal-pruning
- https://odajournal.com/2019/01/23/maximizing-the-accuracy-of-a-cart-tree-model-predicting-missing-data
Did you do your homework on understanding various classification algorithms, their evaluation matrices, cross-validation approach to tune the parameters or to compare those methods, ensemble methods etc.?
I would say first get the basics, then learn the tool.
https://pdfs.semanticscholar.org/68d4/6b8f40ee907a60af9dc9b0b6cac0f5d7a7c9.pdf
- Badges
- Science topic