I compiled/installed the GPU variation of LightGBM and I am able to run the regression example.
Now I would like to use another dataset for regression. How do I specified which column is the target and which ones are the predictors?
Asked
Active
Viewed 146 times
0
-
Which binding (=language) do you use and what interface (python has 2 wrappers) – Mischa Lisovyi May 10 '18 at 13:27
-
Well, in the context of this question: none as I wanted to use the executable through the command line and configuration file. Although, I tried that because I could not find an easy way to use the GPU version of LightGBM through R. – call me Steve May 14 '18 at 17:25
-
if you use the binary directly, then I suppose this is relevant: `Label is the data of first column, and there is no header in the file.` [from the main docs](https://lightgbm.readthedocs.io/en/latest/Quick-Start.html#training-data-format). But using lightgbm from some language is much more comfortable. Did you try setting `device` parameter? – Mischa Lisovyi May 14 '18 at 18:56
-
Cheers! I've been through that line a few time but I was in a "regression" mindset and didn't realised that by Label it meant the Target... – call me Steve May 14 '18 at 23:29