Please, I need clarification on something.
Let's assume I have a dataset from two different website
1. Jaka website
2. Maka website
The first website has a different dataset that contains different information.
Let's assume the Jaka website has a dataset named blood dataset, hands-dataset, legs-dataset, and eyes dataset. Let's assume the Maka website also has some datasets called fingers-dataset, stomach-dataset, eyeball-dataset, toes-dataset, and kneel-datasets.
Let's assume that the largest of these datasets is fingers-dataset from the Maka website.
I used only this fingers-dataset to train LINEARSVC and saved the model.
I am then using the same model to predict the outcome of all other datasets.
The model is performing differently on all Jaka website's datasets, but giving the same result on the Maka website's datasets. I am solving this using a regression method.
The problem here is that I am getting the same result for all the datasets on Maka's website. I am thinking maybe it is because I used one of the datasets on that website to train the model.
Please, am I missing something here?
Linear kernel SVC and Logistic Regression can produce the same decision boundary, that is the exact same model, and therefore the exact same performance metrics, despite using entirely different methods. This can occur when all features are binary, the target variable is binary, and the regularisation parameter for LR is set to zero.
- Badges
- Science topic