predicting the targets for more than one dataset.

[mahedeeb]

When giving datasets for prediting the target defined under the "predict" key in the hyperparameters, predict_unseen_data overwrite old predictions when we have more than one dataset in the predicted key inside the parameters. This means, it is possible to give one dataset for predicting the target

To Reproduce
parameters = {
"data": {
"train": {"features": train_dataframe, "target": train_target.to_numpy()},
},
"split": {
"method": "kfold", # "method":"kfold"
"fold_nr": 5, # fold_nr:5 , "split_ratios": 0.3 # "split_ratios":(0.3,0.2)
},
"model": {"type": "Ridge linear regression",
"hyperparameters": {"alpha": "optimize", # alpha:optimize
},
},
"metrics": ["r2_score"],
"predict": {
"test1": {"features": test_dataframe},
"test2": {"features": test_dataframe},
}
}

Expected behavior
We should get the labels for datasets test1, test2

Desktop (please complete the following information):

• OS: any

Additional context
In the current code, the results from test2 will replace the results from test 1 when running the code
model_index_list, save_models_dir, y_test = flow.training(parameters)
i.e we get only y_test