# Tag Info

Accepted

### Regression At Scale: Best Practices Around Ensuring Quality of a Large Numbers of Forecasts

A few thoughts: If your data is partially hierarchical, then do consider hierarchical forecasting with the optimal reconciliation approach. This very often improves forecasts on all levels. The ...
• 122k
Accepted

### Different result between ARIMA and auto.arima

From the auto.arima documentation: The default arguments are designed for rapid estimation of models for many time series. If you are analysing just one time ...
• 7,743

### Model selection and estimation for pseudo out-of-sample forecasting

The whole idea of pseudo out-of-sample forecasting is that each forecast must be constructed without using any information that would not have been available at the time. This means that you would ...
• 123k

### What is the natural progression from discrete AR models into continuous time?

I disagree that we are using outdated information to predict the future. At time $t$, if you want to predict $y_{t+30}$, you cannot use information from $t+1,t+2,\dots,t+29$ because you do not have it....
• 66.6k

### Prediction Intervals and Alternatives to NHST for testing forecast accuracy between models

There are two separate issues in your question. Prediction intervals: no, narrower intervals are usually not better. After all, we are aiming at a specified coverage. If your PI has the correct ...
• 122k

### Can tbats forecast partial years

TBATS models seasonality in the data with periods equal to the seasonal periods specified. If your data is not periodic, then it's not going to find the seasonal patterns, and may just settle for a ...
• 56.4k

### Incorporating ARIMA errors in a linear regression model

Yes I think you are right, the errors have the AR(1) shape https://otexts.com/fpp2/regarima.html It might be better to compare the in sample or out of sample error metrics
• 283
1 vote

### Incorporating ARIMA errors in a linear regression model

Yes, your interpretation is completely correct. This is absolutely a valid way to go about this. Alternatively, you can throw everything at once into ...
• 122k
1 vote

### Forecasting excess mortality with ARIMA model

If auto.arima handles excess rates $\Delta x_t$ and produces interval forecasts, it will trivially handle cumulative excess rates $x_t$ (by first-differencing them),...
• 66.6k
1 vote
Accepted

### Comparing forecasts under overlapping rolling evaluation windows

For scalar-valued prediction losses, the Diebold-Mariano test is capable of handling multi-step ahead overlapping forecasts the errors (and losses) of which are autocorrelated by design. See Harvey et ...
• 66.6k
1 vote

### On unbiasedness of an optimal forecast

Bias and unbiasedness are concepts that apply to estimates, or estimators, of an unknown population parameter. Thus, I would argue it makes no sense to discuss "the (un)biasedness of a point ...
• 122k
1 vote
Accepted

### Diebold-Mariano test for one-step forecasts using Mean Absolute Scaled Error

The Diebold-Mariano (DM) test is about the expected value of loss from one forecast vs. another forecast, $\mathbb{E}[L(e_1)]$ vs. $\mathbb{E}[L(e_2)]$. The $H_0$ is \$\mathbb{E}[L(e_1)]=\mathbb{E}[L(...
• 66.6k
1 vote

### Approach to Handling Stationarity in Multi-dimensional Time Series Forecasting with AutoARIMA

I think, you could try to use MLForecast library. Maybe StatsForecast could do what you want. Give a chance to that lib. https://nixtlaverse.nixtla.io/mlforecast/index.html It's possible to preprocess ...
1 vote
Accepted

### Time series model selection based on demand

Your question is rather broad. I will give a few thoughts off the top of my head. First off, I recommend this forecasting workflow. (Little surprise there, since I wrote it up.) Decide what kind of ...
• 122k
1 vote

### Problems with time series prediction

The method mentioned by Stephan Kolassa is called Additive Log-Ratio Transformation (alr). For an overview of compositional data analysis, see my answer to R - multinomial logistic regression with ...
• 1,454

Only top scored, non community-wiki answers of a minimum length are eligible