ap.add_argument(
"-d",
"--dataset",
default=
r"C:\Users\matte\Dropbox\fisica_medica\lavori_ieo\ml\radimetrics_train.csv"
)
arg = vars(ap.parse_args())
os.makedirs(r"esperimenti\plain_rf", exist_ok=True)
report = open(r"esperimenti\plain_rf\finetuned_cv_rf", "w")
#load data
print(f"[INFO] Reading data from {arg['dataset']}")
X, y = data_to_model(pd.read_csv(arg["dataset"]))
## PLAIN RANDOM FOREST
report.write("ESPERIMENTO 3. FINE TUNED RANDOMFOREST REGRESSOR:\n")
report.write("\t\t Dati non riscalati, best paramaters\n\n")
scoring = {
'r2': 'r2',
"explained_variance_score": 'explained_variance',
"max error": 'max_error'
}
#scoring=make_scorer(explained_variance_score,max_error,mean_absolute_error,r2_score)
regr = RandomForestRegressor(max_features='sqrt', n_estimators=1000)
scores = cross_validate(regr, X, y, cv=10, n_jobs=-1, verbose=1)
if report is not None:
report.write(f"[RESULT] mean_absolute_error score: {mae} mSv\n")
report.write(f"[RESULT] max_error score: {max_er} mSv\n")
report.write(f"[RESULT] mean_percentage_error score: {mape} %\n")
report.write(f"[RESULT] mean_percentage_error biased score: {mape_b} %")
return mae,max_er,mape,mape_b
### CARICO TUTTI I MODELLI FACCIO LE PREVISIONI E SPUTO I RISULTATI
# load the model from disk
#per quelli che usano dati riscalati
X,y=data_to_model(original_data,sep="/")
print(f"[INFO] final dimension of dataset X: {X.shape}\ty:{y.shape}")
mm=MinMaxScaler()
X_norm=mm.fit_transform(X)
### RANDOM FOREST
report=open("physico_test_report_nothr", "w")
print(f"\t\t[RANDOM FOREST]\n")
filename=r"esperimenti\plain_rf\plain_rf.sav"
loaded_model = pickle.load(open(filename, 'rb'))
