Modelling Trend

This is a minimalistic example of trend modelling in Neuralprophet by defining changepoints.

m = NeuralProphet(
    n_changepoints=100,
    trend_reg=2, 
    yearly_seasonality=False,
    weekly_seasonality=False,
    daily_seasonality=False,
)
metrics = m.fit(df, freq="D")
forecast = m.predict(df)

The components plot looks like below with only trend and residuals as a components.

plot-comp-1{: style=”height:400px”}

The coefficients plot should show the coefficients corresponding to the 100 changepoints.

plot-param-1{: style=”height:400px”}