The predicted values are not in the expected patterns.

[saksonita]

Dear NeuralProphet team,

NeuralProphet has a really good performance compared to others. However, after I observed the predicted values, those values doesn't follow the original pattern of the datasets. In my case, as you may see in the figure below.

However, the predicted value doesn't provide the same patterns.

Below are my code used to define the model.

#Dependence Variables
col_lst = ['daywk', 'tempAvg', 'tempHigh', 'tempLow', 'humAvg', 'rain',
       'insolAmt', 'tempground', 'moisture', 'y_mean_lag6', 'y_mean_lag14',
       'y_mean_lag28', 'y_std_lag6', 'y_std_lag14', 'y_std_lag28']


m = NeuralProphet(
    n_forecasts=365,
    n_lags=7,
    weekly_seasonality=7,
    yearly_seasonality='auto',#주간 계절성 설정
    d_hidden=128, #은닉층 뉴런 설정
    learning_rate=0.02, #학습률 설정
    batch_size=30, #배치 사이즈 설정
    epochs=100, #학습 횟수
)

m = m.add_lagged_regressor(names=col_lst, normalize="soft")

I am looking forward to your response.

Best Regards,
saksonita