Model

Hyper-parameters

VAR

lag = 2

SVR

kernel = “rbf”, gamma = 0.1, C = 120

RF

n estimators = 30, max depth = 7

LSTM

epochs = 500, batch size = 10, optimizer = “adam”, n past = 7

Layers: LSTM1(128), LSTM2(64), Dropout(0.5), Dense(1)