通过Statsmodels进行时间序列预测

Question

I used the following code the make some predictions :

Just to give your a quick look at the dataset (df.head()):

  places_occupees                date
0              238 2017-01-01 00:00:00
1              238 2017-01-01 00:01:00
2              238 2017-01-01 00:02:00
3              238 2017-01-01 00:03:00
4              238 2017-01-01 00:04:00

Then I make predictions for the 7 days to come:

X = places.values
train, test = X[1:len(X)-7], X[len(X)-7:]
# On entraîne le modèle d'autoregression
model = AR(train)
model_fit = model.fit()
print('Lag: %s' % model_fit.k_ar)
print('Coefficients: %s' % model_fit.params)
#On effectue des predictions
predictions = model_fit.predict(start=len(train), end=len(train)+len(test)-1, dynamic=False)

I get the following output:

This is how the plot looks like:

But what I really want to do is to make some predictions for the next 24 hours and not for the next 7 days. How can I do that? How shoud I modify the code above?

Thanks in advance.

Answer 1

您应该编码

train, test = X[:-30], X[-30:]