def build_regression_model(args):
settings.NPARA = args.npara
settings.PROBACUTOFF = args.probacutoff
#settings.SAVEMODEL=args.savemodel
settings.SAVEPRED = args.savepred
#settings.MICROSPEC=args.microspec
#settings.GRIDSEARCH=args.gridsearch
settings.LATENT = args.latent
X1, X2, Y1, Y2, O1, O2, V = load_datasets(training=settings.FIT,
test=settings.PREDICT,
y_name=settings.ACTIVITY)
if settings.VERBOSE == 1:
print(
"\nTRAINING SET:\nN objects = %s\nN independent vars = %s\nN dependent vars: 1 (%s)\n\nTEST SET:\nN objects = %s\n"
% (len(O1), len(X1[0]), settings.ACTIVITY, len(O2)))
if settings.MODEL == "PLS":
y_train_pred, y_test_pred = run_pls(X_train=X1,
X_test=X2,
y_train=Y1,
y_test=Y2)
else:
Y1_pred, Y2_pred, Y1_prob, Y2_prob = modelling(
X_train=X1,
X_test=X2,
Y_train=Y1,
model_type=settings.MODEL,
nondef_params=settings.NPARA,
sm=settings.SAVEMODEL,
mc=settings.MULTICLASS)
def build_auto(args):
X1, X2, Y1, Y2, O1, O2, V = load_datasets(training=settings.FIT, test=settings.PREDICT, response=settings.RESPONSE)
#settings.VAR_NAMES=V[1:-1]
#settings.VARS=V
run_procedure(X1=X1, X2=X2, Y1=Y1, Y2=Y2, O1=O1, O2=O2)
def build_class_regression_model(args):
settings.LATENT = args.latent
settings.HIGHTHRESHOLD = args.highthreshold
settings.LOWTHRESHOLD = args.lowthreshold
settings.SAVEPRED = args.savepred
settings.VERBOSE = -1
X1, X2, Y1, Y2, O1, O2, V = load_datasets(training=settings.FIT,
test=settings.PREDICT,
y_name=settings.ACTIVITY)
if settings.MODEL == 'PLS':
run_pls(X_train=X1,
X_test=X2,
Y_train=Y1,
Y_test=Y2,
O_train=O1,
O_test=O2,
nlv=settings.LATENT,
lt=settings.LOWTHRESHOLD,
ht=settings.HIGHTHRESHOLD)
def build_classification_model(args):
settings.NPARA = args.npara
settings.MULTICLASS = args.multiclass
settings.PROBACUTOFF = args.probacutoff
settings.SAVEMODEL = args.savemodel
settings.SAVEPRED = args.savepred
settings.GRIDSEARCH = args.gridsearch
settings.BACKFEEL = args.backfeel
X1, X2, Y1, Y2, O1, O2, V = load_datasets(training=settings.FIT,
test=settings.PREDICT,
y_name=settings.ACTIVITY)
settings.VAR_NAMES = V[1:-1]
if settings.GRIDSEARCH:
gridsearchcv(X1=X1,
Y1=Y1,
X2=X2,
Y2=Y2,
grid=param_grids[settings.MODEL])
else:
make_one_model(X1=X1, X2=X2, Y1=Y1, Y2=Y2, O1=O1, O2=O2)
def run_dmody_regression_operations(args):
settings.NPARA, variables.DMODY = False, True
variables.X_tra, variables.Y_tra, variables.O_list, variables.V_list = load_datasets(
training=settings.FIT, response=settings.RESPONSE)
print(
"\nTRAINING SET:\nN objects = %s\nN independent vars = %s\nN dependent vars: 1 (%s)"
% (len(variables.O_list), len(variables.X_tra[0]), settings.RESPONSE))
df_results = pd.DataFrame({'Y_exp': variables.Y_tra},
index=variables.O_list)
for a in algorithms_list:
settings.MODEL = a
print("\nPerforming %s+LOO..." % a)
run_model_training()
distances = calc_dmody()
df_results['Y_pred (%s)' % settings.MODEL] = variables.Y_pred
df_results['DModY (%s)' % settings.MODEL] = distances
#print(df_results)
df_results.to_csv("DModY.csv", sep=";")