❓ FAQ¶
Q: I have a clinical dataset with some patient data. Can I use this data with CliMB?
A: This depends on the nature and requirements of your data. Reading the 🔏 Data Privacy section of this documentation should allow you to make an informed decision on this.
Q: My dataset has so-and-so many variables and so-and-so many records (e.g. patients). Will a dataset of this size work with CliMB?
A: This depends on your hardware configuration and the details of your dataset(note: for minimal hardware requirements, please refer to the 📦 Installation).
If your dataset has more than a few 10s of features, the data exploration step may produce too much output, and use up the LLM context window. We recommend using datasets with no more than ~50 features.
A larger number of records (rows of data) will lead to a longer execution time of various tools, especially the predictive modelling steps and the feature importance steps. For a modern workstation with a GPU, please use the following very rough guide for time estimate: 5 minutes per 1,000 records for the predictive modelling step, and up to 5 times slower for the feature importance step.
Q: I have problems when installing or running CliMB. Where can I get help?