New answers tagged forecasting
1
vote
How to determine the appropriate forecasting techniques for non-stationary univariate time series data?
For a simple illustrative case, take the CO2 data, pretty linear, so a simple parsimonious model, even a simple linear trend model, or better a simple one parameter auto-regressive time series model, ...
0
votes
Is there anyway I can limit my forecast to a upper limit using bsts package?
As with any other time series method you can forecast a transformation of $z = g(y)$ instead of $y$ directly. Such that the inverse of the transformation is bounded.
If your example has upper $b$ and ...
0
votes
Learning stochastic pattern using RNN
No, it is not possible since standard neural networks give you just point estimates. If you want to learn a distribution you should look into probabalistic deep learning.
1
vote
Accepted
Long-term trend prediction of time series data
I guess your time series have some kind of seasonality, since it is a product which is probably sold more often during some periods of the year or month?
A way to tell more about the trend is using ...
0
votes
Accepted
Volatility forecast using ARIMA GARCH
Regardless of the conditional mean model (constant, ARMA or something else), GARCH is a model for the conditional variance of the variable of interest (e.g. return on exchange rate).* Combined with a ...
0
votes
Using Confidence Intervals for finding confidence of forecasting model
The basic question is correct - as a forecast goes farther out in time, the confidence intervals become wider - sometimes extremely fast - thus reducing the confidence in the conclusions provided by ...
1
vote
Accepted
Time Series is not white noise but Random Walk
To proceed with this series for building forecasting , I read many places if your data is not white noise and does not have random walk then you can model forecasting , Otherwise if your time series ...
0
votes
Arima or Sarima model parameters
Don't fit an ARIMA model by hand. It's much better to rely on a tested software implementation, like the forecast and fable ...
0
votes
How to improve the accuracy of an ARIMA model
I have never seen a time series with negative values where a MAPE would make sense.
I would very much recommend you disregard the MAPE. See What are the shortcomings of the Mean Absolute Percentage ...
0
votes
Giving more importance to under prediction (mean absolute error) than over prediction for forecasting
You should ideally look up the economics of the situation involved.
What's the actual financial cost of being understaffed by one? Is it a schedule slippage of one month, which cases late entry to ...
0
votes
Accepted
Giving more importance to under prediction (mean absolute error) than over prediction for forecasting
The good news: if you use the MAE, chances are good you are already under-predicting. Specifically, the MAE elicits the conditional median, not the mean, and if your series has a skewed conditional ...
1
vote
Steps for improving NegBinom regression model
I personally prefer working with R, which has much better facilities for time series forecasting. First off, your series has such high counts that count models are probably not going to give you much ...
6
votes
Accepted
Why is my MASE systematically equal to 1?
Let $y_t$ be the actual and $\hat{y}_t$ the forecast at time $t$, for times $t=1, \dots, T$.
You calculate the denominator used in the MASE as
$$ d := \frac{1}{T}\sum_{t=1}^T|\hat{y}_t-y_t|.$$
For a ...
1
vote
Accepted
Forecasting model for small dataset with many parameters?
If the other predictors presumably influence your $y$, but not vice versa, then a lasso might be useful. The problem then is that you would need to forecast your predictors themselves.
If your $y$ ...
0
votes
Accuracy of point forecasts vs. average accuracy of multistep forecasts?
This is a great question, and I think it is telling it does not have any answers (even though that might be down to the title that is bit misleading I think).
I always found it puzzling that the ...
0
votes
Accepted
Forecasting avg. flight occupancy
It sounds like you have daily data, based on your asking about leap years and Christmas.
A sensible baseline is a time series model. For daily data, you may have multiple-seasonalities, specifically ...
7
votes
ARIMA(0,1,0) predictions are previous values from one year before
As frank writes, predicting last year's data point is exactly what an I(1) model is all about, so there is no reason for concern here.
ARIMA modeling is quite non-trivial, and the old Box-Jenkins ...
4
votes
ARIMA(0,1,0) predictions are previous values from one year before
The dickey-fuller test shows that the data is stationary and looking at the acf and pacf I decided to use an ARIMA (0,1,0) to model the data.*
This is a strange decision, as your ADF test result ...
4
votes
ARIMA(0,1,0) predictions are previous values from one year before
The ARIMA(0, 1, 0) model is just the I(1) model, which is:
$$
y_{t+1} = y_t + \epsilon_{t+1},
$$
also called the random walk. As you can see, the expectation of $y_{t+1}$ is then just the previous ...
2
votes
Time series analysis if it is predictable or not
Both are predictable, just not equally well. Consider a simple nonstationary random walk
$$
y_t=y_{t-1}+\epsilon_t=y_0+\sum_{j=1}^t\epsilon_j
$$
Suppose we observe this process until $T$ and want to ...
0
votes
Forecasting time series with multiple seasonaliy by using auto_arima(SARIMAX) and Fourier terms
A bit late, but I had a couple ideas you might try:
Specifying the arima model parameters instead of asking auto_arima to find. I was largely unimpressed with auto_arima since it calibrates based on ...
1
vote
How to set limits on STLF forecast in R
+1 to Rob's answer.
To your second question: yes, the RMSE is probably precisely the error measure you want. It will be minimized in expectation by an unbiased prediction of the probability of rain. ...
2
votes
Accepted
How to set limits on STLF forecast in R
You're applying a method designed for a continuous distribution to a binary outcome. No amount of adjusting the method will fix that. Use a method designed for a binary outcome instead. e.g., a ...
1
vote
Usage of SVR in Non Stationary Time Series Forecasting
The condition of independent identical distributions (iid) refers to the $(x,y)$ data points. Independence means:
$$
p((x_i, y_i), (x_k, y_k)) = p((x_i, y_i)) p((x_k, y_k)),
$$
i.e. the sampling of $(...
Top 50 recent answers are included
Related Tags
forecasting × 3625time-series × 2227
arima × 915
r × 887
predictive-models × 372
regression × 338
machine-learning × 241
seasonality × 176
neural-networks × 139
exponential-smoothing × 131
python × 127
prediction-interval × 107
garch × 86
vector-autoregression × 82
accuracy × 80
autocorrelation × 72
stationarity × 72
model-selection × 65
autoregressive × 65
cross-validation × 64
econometrics × 58
multiple-regression × 57
model-evaluation × 56
modeling × 54
lstm × 54