Hi, sorry to bother, I am new to Knime, and I am doing this project that I’m supposed to analyse the data, and extract the most relevant information. I’m testing the loan status results, but I can’t get the acc up, since I’m new, i am not sure what to do to improve it. So I am asking for some advice.
KNIME_project.knwf (260.9 KB)
loan_test.csv (530.5 KB)
I can’t upload the train dataset because its too big
Hi @J_Pinto1,
Welcome to the Forum.
It looks like the current approach treats this as a binary classification problem (Approved vs Rejected), although the dataset contains a third class: Pending. Consider whether Pending should be included as a separate class. If it represents an ambiguous or transitional state, removing or reassigning it may help simplify the task.
Additionally, avoid normalizing the entire dataset before splitting. This can lead to data leakage. Instead, split the data first, then normalize using parameters derived from the training set (after X-Partitioner).
Finally, consider tuning the model’s hyperparameters. Adjusting values such as split criteria, tree depth, or number of models—depending on the algorithm—can help improve accuracy.
Best,
Keerthan