Questions tagged [forecasting]

Prediction of the future events. It is a special case of [prediction], in the context of [time-series].

Filter by
Sorted by
Tagged with
0 votes
0 answers
3 views

Paradox of Brier skill score of perfectly calibrated output?

Given outcomes, $y \in \{0,1\}$ and outputs $o = f(x) \in \mathbb R, o \in [0,1]$, I'm interested in the case where the model $f$ perfectly models the variable $Y$. Since $Y$ is Bernoulli, this means $...
Firebug's user avatar
  • 19.1k
0 votes
0 answers
21 views

Predict COVID spread using latitude and longitude and time

I have a data that has latitude and longitude of individuals and the timestamp of geographical locations. I want to predict the spread of COVID using R using latitude/longitude and the time as well. I ...
William's user avatar
  • 131
0 votes
1 answer
14 views

Why the differenced at lag 12 time series of a SARIMA(0,0,0)(0,1,1)_12 model follow the MA(1) pattern with step 12?

I am trying to understand why the ACF of the seasonally differenced series reveals the AR of MA structure of the original series. For example: The following lines creates a SARIMA(0,0,0)(0,1,1)_12 ...
Epameinondas's user avatar
0 votes
0 answers
21 views

What is the ARMAX model specification of the following economic setting?

I am currently doing a project estimating electricity prices in France, however, my modelling skills are lacking. I have hourly data on spot prices, which are determined per separate hour, one day ...
Zillah's user avatar
  • 31
2 votes
1 answer
41 views

Contradictory Sources on Seasonality being a nonstationarity

I have been trying to figure out whether seasonality means nonstationarity, and the answers from many (often reliable) sources seem to be contradicting. (lets define stationarity as weakly ...
da7666's user avatar
  • 31
0 votes
0 answers
10 views

Modeling non-negative time series with square root decay?

Q: How should I model a non-negative time series $y_t$ which exhibits square-root decay? More specifically, a time series $y_t$ whose square-root differences $\sqrt{y_t}-\sqrt{y_{t-1}}$ are linear and ...
lowndrul's user avatar
  • 2,127
1 vote
1 answer
11 views

Rolling forecasts where horizon is larger than step-size

Is it bad practice to perform rolling time-series forecasting where the forecast horizon is greater than the step-size? For example, if I have a model which produces weekly forecasts on a rolling day-...
ron burgundy's user avatar
0 votes
0 answers
29 views

Why do we use only event history for temporal point processes?

I'm trying to understand temporal points processes. In particular, neural TPPs. In all the works I've read, the only features fed into the model are a sequence of event timestamps and marks if they ...
JulioHC00's user avatar
0 votes
0 answers
10 views

Can we use decomposed time series in a test set to obtain models accuracy metrics?

Currently I'm cross validating (rolling time window, different test lengths) several forecast models to obtain performance comparison on highly skewed time series (due to true ocassional outliers). My ...
Tom's user avatar
  • 481
0 votes
0 answers
19 views

How to determine statistical significance for a time series and forecasts?

With a simple example of mortality rates, and a basic three-year mean baseline: ...
electronix384128's user avatar
4 votes
2 answers
40 views

Forecasting a series that comes with uncertainty

I have a time series resulting from a spatiotemporal aggregation on the spatial domain. As a result, I have a central measurement (let's say mean average) and a dispersion (let's say standard ...
Ricardo Barros Lourenço's user avatar
6 votes
2 answers
401 views

Re-selecting best sales forecasting model each month. Is this overfitting?

One of our teams works on sales prediction, and they run ~10 models each month for each product (+100). Then they use the best fitting model for next period prediction, which may be a totally ...
Luis's user avatar
  • 595
0 votes
0 answers
18 views

Different results when fitting ARIMA model for levels vs ARMA model for first differences in R

In the following code I show that I get different forecasts when fitting an ARIMA(2,1,0) for cumulative sums of a generated AR(2) model vs. fitting an ARMA(2,0) for the AR(2) itself. Can anyone point ...
Mr Frog's user avatar
  • 339
1 vote
1 answer
19 views

How to deal with zeros when using NBEATS to forcast demand?

So, I'm doing forecasting demand for a hotel using NBEATS, which is hierarchically organized. However, I'm facing an issue with the time series data at the bottom (room numbers), as they're filled ...
Jakov Gl.'s user avatar
  • 111
0 votes
0 answers
30 views

Forecasting RNN and LSTM without X_test

Dear StackExchange Community, My data is composed of only 1 time series variable (Stock prices of an asset) I have splitted it to train and test subsets. I have tarined an RNN and LSTM models with ...
Amirgiano's user avatar
0 votes
0 answers
17 views

A good variable to make a regression model for gas usage over the years for a city

so I'm new to statistics, I'm trying to make a regression model in Excel, explaining why, or due to what variable, does the gas usage change over the years. I tried using a basic Y variable - Time - ...
Nero's user avatar
  • 1
1 vote
0 answers
16 views

Is the ETS function from forecast appropriate for skewed environmental time series data? does the time-series data need to meet certain criterion? [closed]

I have environmental data that is usually just analysed with the nonparametric man-kendall and sens slope, because it is skewed. Can the ETS function which picks what exponential smoothing model to ...
Justin Murphy's user avatar
2 votes
0 answers
73 views

Issue with SARIMA model for PM10 concentration forecasting with m=365 [migrated]

I'm trying to build a SARIMA (Seasonal Autoregressive Integrated Moving Average) model for forecasting PM10 concentrations based on five years of data. However, when I set the seasonal parameter m to ...
Divyansh Sharma's user avatar
1 vote
0 answers
23 views

Is stationarity important when using boosting models?

I've studied time series for the past months and I've seen mainly two ways of building a forecasting model: Using ensemble algorithms and making the time series look like a cross-sectional data, in ...
trder's user avatar
  • 660
2 votes
1 answer
37 views

Seasonal Time Series Forecasting and MASE

I am creating a a number of forecasts for sales at various levels of data aggregation, based on the properties of the products (eg. is it in a bottle or a can). I plan to create multiple models and ...
ForecastNoob's user avatar
0 votes
0 answers
50 views

How much statistics knowledge is enough for studying time series?

I am interested in studying time series (both theoretical and applied), my background is in mathematics and my probability knowledge is well enough but I haven't studied stats deeply (only one course ...
Pat's user avatar
  • 1
1 vote
0 answers
24 views

ETS Confidence Intervals in R are several orders of magnitude larger than the time series itself?

I am forecasting a time series with confidence intervals using the ets model in R. Here is the time series: Running the following R code: ...
Justin Furlotte's user avatar
1 vote
1 answer
59 views

Violation of assumptions in Generalized additive models (GAMs)

I wanted to forecast Dengue count for the given weather data covariates: MeanT Mean temperature MeanRH Mean relative humidity MeanSM Mean soil moisture MeanVP Mean vapor pressure MeanWS Mean ...
Quantam's user avatar
  • 23
2 votes
1 answer
77 views

Which Forecast Evaluation Metric To Use?

It is a forecasting problem. I need an evaluation metric which penalizes under-predictions more than over-predictions. Also I want it's range in certain interval (say 0-100), so that it becomes easier ...
Shardul Pingale's user avatar
1 vote
0 answers
11 views

Is there a correction for samples from a (linear) Prophet model when trained on an inhomogenous Poisson point process?

Facebook's Prophet is a popular modelling choice for time series forecasting in production due to many steps being automated (and thus convenient). This can sometimes lead to over-reliance on it when ...
Galen's user avatar
  • 8,502
0 votes
0 answers
18 views

can I use causalimpact (BSTS) to forecast without an intervention?

As a data scientist without much formal training, I'm looking to get some professional feedback on the following question: Is it ever advisable to use causalimpact (or I guess BSTS generally) to ...
ouonomos's user avatar
  • 121
2 votes
2 answers
83 views

Incorporating ARIMA errors in a linear regression model

I am working with economic data and trying to create a linear regression model for forecasting purposes. The dependent variable data is in terms of percentage change and I've differenced the ...
Amy K's user avatar
  • 151
3 votes
1 answer
41 views

Forecasting excess mortality with ARIMA model

I am using the forecast package by Prof Hyndman, and have had success fitting ARIMA models to excess mortality (from the COVID-19 pandemic) data. I am currently trying to produce plots for cumulative ...
Jina A.'s user avatar
  • 31
1 vote
0 answers
31 views

Multilayer Perceptron vs. Recurrent Neural Network for Time Series Forecasting: Utilizing Multiple Lagged Values

I am currently analyzing daily sales data for a product sold across multiple stores using a Multilayer Perceptron (MLP) model. For simplicity, let's assume it consists of a single layer, structured as ...
Girigio's user avatar
  • 63
3 votes
1 answer
49 views

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

Here Rob Hyndman says In the predictive approach to statistics (McLean, 2000), problems of statistical analysis are viewed as prediction problems, and the central theme is a statistical (probability) ...
pandashelp's user avatar
7 votes
1 answer
65 views

On unbiasedness of an optimal forecast

Diebold "Forecasting in Economics, Business, Finance and Beyond" (v. 1 August 2017) section 10.1 lists absolute standards for point forecasts, with the first one being unbiasedness: Optimal ...
Richard Hardy's user avatar
0 votes
0 answers
14 views

Extrapolating/forecasting treatment effects from difference-in-difference model

My goal is to extrapolate or forecast dynamic treatment effects into the future using a fitted model. My data consists of two groups (treated and control), seven time points, and a continuous outcome. ...
CaptainAardvark's user avatar
2 votes
1 answer
34 views

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

This is my first time doing time series forecasting, so I am sorry for any inconsistencies in my question. But I have two different models that I want to compare. On Wikipedia, I read about Mean ...
Oskar Weber's user avatar
0 votes
0 answers
11 views

weekly time series forecasting with exogenous variables retrain cadence

I have a weekly forecasting model, which has exogenous variables. The forecasting horizon is 8 weeks ahead - 8 data points. I do forecast (8 weeks) the exogenous variables to use them as input for ...
janicebaratheon's user avatar
0 votes
1 answer
28 views

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

I am working on a time series forecasting project for a meal delivery service that operates in multiple cities. The company has several fulfillment centers across these cities for dispatching meal ...
user172500's user avatar
2 votes
1 answer
30 views

Can tbats forecast partial years [closed]

I have several years of daily data from August to December. I read that I can use tbats from the forecast package. Each year has approximately 153 days but ...
Salvador's user avatar
  • 207
3 votes
0 answers
37 views

What critical level to use in diagnostic tests for model selection in forecasting?

I have been reading Hyndman & Athanasopoulos "Forecasting: Principles and Practice" (newest edition here) recently, and I noticed something that I regard as a possible inconsistency. On ...
Richard Hardy's user avatar
0 votes
0 answers
29 views

In prediction modelling, is it bad practice to combine differing sampling frequencies of covariates into the same model?

An example of this type of prediction task is modelling economic time-series data. Depending on the type of data being reported, the sampling frequency varies: GDP is reported quarterly, employment ...
ron burgundy's user avatar
0 votes
0 answers
14 views

Approach for forecasting time-series with several time-series of different frequency

The time-series I'm looking to forecast consists of daily data. The other time-series are made up of each of daily, weekly and monthly levels. Top of mind approach is to use an MV regression, ...
Chris's user avatar
  • 101
1 vote
1 answer
54 views

Comparing forecasts under overlapping rolling evaluation windows

Two models with differing assumptions each purport to provide a forecast for a vector of values 1 year in the future. Each model has been run at the start of each month for three years: Run Date ...
David's user avatar
  • 515
0 votes
0 answers
16 views

How to Evaluate a Single-Value Prediction for a 6-Month Period Against Historical Data?

I'm tackling a time-series forecasting issue with daily granularity, aiming to predict a single aggregate value that represents the total sum of incidents over a 6-month period. My approach involves ...
Amit S's user avatar
  • 57
0 votes
0 answers
25 views

Can SHAP values across different models in the same family of models trained on different datasets be aggregated?

Let's say I have time - series data for 100 products in a particular store. I fit 100 regression models to generate 1-step forecasts for these products. Let's say that all features are common across ...
Tamojit Maiti's user avatar
0 votes
0 answers
30 views

Autoregession meets multiple regression? - Help with verbiage and approach

Needing some help with verbiage and opinions on how I am approaching this model. I have counts of people over the past 24 months. month count 1 100 2 105 ... ... 24 200 First, I reverse the ...
Tyler Brown's user avatar
8 votes
1 answer
288 views

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

Background Often I am forecasting possibly one up to a few dozen variables in a project, but I have an upcoming project that will involve forecasting thousands of variables. I have some ideas of my ...
Galen's user avatar
  • 8,502
2 votes
1 answer
39 views

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

Lets say we want to predict a single target variable and we have 10 regressors/features. Assume we would like to predict 30 days ahead (daily predictions up to 30 days ahead) and our data is a daily ...
MilTom's user avatar
  • 369
2 votes
1 answer
51 views

Time series model selection based on demand

I am working on a time series forecasting project in my office. We have to forecast the demand for 10 products at each store level (no hierarchy). Just at store level is enough..We have 20 stores.So, ...
The Great's user avatar
  • 3,272
0 votes
0 answers
60 views

Inclusion of prior knowledge about forecast to a (Bayesian) time series model

I am trying to construct an inflation forecasting model with external regressors and seasonality. It came to my mind that a very useful information for prediction could be the consensus forecast of ...
PK1998's user avatar
  • 151
4 votes
1 answer
162 views

Different result between ARIMA and auto.arima

I have conducted an analysis on a time series dataset, and the issue is that the 'manual' ARIMA (SARIMA) performed in Gretl gives me a model (3,0,1)(1,1,1) with all *** and AIC=1595.332. However, when ...
Leonard Banković's user avatar
3 votes
1 answer
72 views

Density Forecasts with GAMLSS

Does someone know the function to create density forecasts within the GAMLSS Package? The predict. Formula is not the right one. Predict do Point Forecasts
Nushko's user avatar
  • 31
0 votes
0 answers
13 views

Diebold-Mariano Revisited: what is a reasonable parameter count for information criteria when the model is complex?

Diebold (2015) wrote a follow-up paper/essay reflecting on how his work with Mariano to develop the Diebold-Mariano test has been abused over the years. One of the main points in the follow-up paper ...
Dave's user avatar
  • 62.5k

1
2 3 4 5
78