search

Professor-G / MicroLIA

Gravitational microlensing classification engine using machine learning
GNU General Public License v3.0
13 stars 6 forks source link

TypeError: hyper_opt() got an unexpected keyword argument 'limit_search' #11

Closed eddiemorris135 closed 1 year ago

eddiemorris135 commented 1 year ago

I get the following error when I run through Example: OGLE II

model = models.Classifier(data_x, data_y, clf='rf') model.create() Imputing data... Running feature selection... 0%| | 0/50 [00:00<?, ?it/s] Feature selection complete, 79 selected out of 148! Traceback (most recent call last):

File "C:\Users\edmun\AppData\Local\Temp\ipykernel_6260\41337179.py", line 2, in model.create()

File "C:\ProgramData\Anaconda3\lib\site-packages\MicroLIA\models.py", line 186, in create self.model, self.best_params, self.optimization_results = hyper_opt(data_x, self.data_y, clf=self.clf, n_iter=self.n_iter,

TypeError: hyper_opt() got an unexpected keyword argument 'limit_search'

Professor-G commented 1 year ago

Hi Eddie, I know the cause of this issue, MicroLIA is currently outdated as I continued development for this program on my pyBIA ensemble_model module. I am going to translate the updates to MicroLIA this week and release the new pip version, I will ping you when it's ready. Cheers

eddiemorris135 commented 1 year ago

Cool. Look forward to it. Thanks in advance.

eddiemorris135 commented 1 year ago

Hi Professor-G Any chance to look at the update? Thanks Ed

Professor-G commented 1 year ago

The code has been updated with the models module now re-named to ensemble_model

The optimization routine is sluggish on my machine so for the time being I recommend optimize=False, until version 2 release. Note no pip update has been released, please clone the repo and do pip install .

UPDATE: optimize=True seems to work fine, although i created a training set with only 50 objects per class and record approx one optimization trial per minute.

eddiemorris135 commented 1 year ago

Iv had trouble installing the clone via anaconda (conda install). would a standard pip install on a new machine work to reflect the new update (without anaconda)?