Enrol to start learning
Reading is open to everyone. Enrolling is free, and it is what unlocks the audio lessons, practice tests and progress tracking.
10.6. Seasonal Models: SARIMA and SARIMAX
Interactive Audio Lesson
Unlock the classroom podcast
The transcript is above and free to read. A free account plays the conversation back.
Create a free account
Today, we are going to learn about SARIMA, which stands for Seasonal Autoregressive Integrated Moving Average. Can anyone tell me why we might need a seasonal model?
I think it’s because some data changes with the seasons, like sales during holidays?
Exactly! Seasonal models help us capture these repetitive patterns. SARIMA specifically factors in these seasonal components. Now, what do you think the 'AR' in SARIMA stands for?
It’s Autoregressive, right? That means it uses past values to predict future values?
Correct! Remember, autoregressive means we’re linking back to past observations. We also have 'Integrated,' which helps with non-stationarity. Let's not forget that the seasonal component would be represented with capital letters. Can anyone give me an example of a seasonal cycle?
The seasons of the year! Like sales spikes every December.
Great example! Now let’s summarize what we've learned so far: SARIMA is a vital tool for capturing seasonal patterns in data.
Unlock the classroom podcast
The transcript is above and free to read. A free account plays the conversation back.
Create a free account
Next, let’s explore SARIMAX, an extension of SARIMA. Why do you think we might want to add exogenous variables to our models?
So we can include factors like marketing campaigns or economic indicators that might affect our sales?
Exactly! SARIMAX allows you to incorporate these external influences. It’s defined just like SARIMA but with additional variables. Can someone tell me the significance of including these exogenous variables?
It helps improve our forecasts by considering factors outside the inherent time series data!
Fantastic! Including relevant external factors can enhance predictive accuracy. As a recap, SARIMAX maintains the seasonality structured like SARIMA but allows for more comprehensive forecasting using additional data.
Unlock the classroom podcast
The transcript is above and free to read. A free account plays the conversation back.
Create a free accountNow, let’s compare SARIMA and SARIMAX. Can anyone highlight a key difference between the two?
SARIMA doesn’t include any outside variables, while SARIMAX does?
That’s right! And this also means that SARIMAX is applicable in more diverse forecasting scenarios. Can someone think of a context where SARIMAX would be particularly useful?
In retail, when combining seasonal effects with a promotional campaign!
Exactly! Retailers can use both seasonality and the impacts of their marketing efforts. Remember, the choice between these models often depends on your data and specific forecasting needs. Summarizing, SARIMAX is more versatile due to the inclusion of exogenous variables.
Overview
Short Summary
This section covers SARIMA and SARIMAX, models specifically designed to handle seasonality in time series data, where SARIMA processes seasonal components explicitly and SARIMAX extends it by including exogenous variables.
Medium Summary
The section discusses the Seasonal ARIMA (SARIMA) model, which incorporates seasonal components into time series forecasting, and introduces SARIMAX, an extension that allows for the inclusion of external regressors. Both models are crucial for capturing seasonal patterns effectively in various applications.
Detailed Summary
Seasonal Models: SARIMA and SARIMAX
In time series forecasting, seasonality can significantly affect the patterns in data. The Seasonal Autoregressive Integrated Moving Average (SARIMA) model is specifically designed to handle these repeated patterns over fixed periods (e.g., monthly, quarterly). It incorporates seasonal aspects into the ARIMA framework by adding seasonal terms into the model parameterization. The parameters for SARIMA are expressed as
(p,d,q)(P,D,Q)s, where:
- p: Non-seasonal autoregressive order
- d: Non-seasonal differencing order
- q: Non-seasonal moving average order
- P: Seasonal autoregressive order
- D: Seasonal differencing order
- Q: Seasonal moving average order
- s: Length of the seasonal cycle (e.g., 12 for monthly data)
SARIMAX is an extension of SARIMA that allows for external factors or exogenous variables to be included in the modeling process. This capability enhances the model's forecasting power by leveraging additional relevant information that can influence the target variable. By using SARIMAX, forecasters can make more accurate predictions by taking into account these external variables alongside the seasonal patterns inherent in the time series data.
Reference YouTube Videos
Audio Book
Unlock the audio lesson
The script is above and free to read. A free account plays it back, in the voice you pick.
Create a free accountSARIMA (Seasonal ARIMA)
- Handles seasonality explicitly.
- Parameters: (p,d,q)(P,D,Q)s
- Where capital letters denote seasonal components, and 𝑠 is the length of seasonality (e.g., 12 for monthly data).
Detailed Explanation
SARIMA models are designed to account for the seasonal patterns present in time series data. Unlike regular ARIMA models, which focus only on non-seasonal characteristics, SARIMA explicitly incorporates the seasonal aspects by utilizing both seasonal and non-seasonal parameters. The parameters are divided into two types: the non-seasonal parameters (p, d, q) and the seasonal parameters (P, D, Q). The 's' refers to the seasonal frequency, which indicates how many periods are in a season, such as 12 for monthly data where each year comprises 12 months.
Examples & Analogies
Think of SARIMA like preparing for a seasonal festival. Just as a festival planner must consider various elements like decorations (seasonal parameters) and logistics (non-seasonal parameters) to ensure the event runs smoothly, SARIMA combines these different factors to predict future values, considering both long-term trends and regular seasonal patterns.
Unlock the audio lesson
The script is above and free to read. A free account plays it back, in the voice you pick.
Create a free accountSARIMAX
- Extension of SARIMA that includes exogenous variables (regressors).
Detailed Explanation
SARIMAX stands for Seasonal Autoregressive Integrated Moving Average with eXogenous inputs. This model extends the SARIMA framework by integrating exogenous variables into the analysis. Exogenous variables are external factors that can influence the time series being predicted. For instance, if forecasting airline passenger numbers, variables such as ticket prices, economic indicators, or seasonal events could be included as regressors to enhance the model's accuracy.
Examples & Analogies
Imagine you are predicting the sales of ice cream. Besides considering the historical sales data (like previous years' sales trends and seasonal peaks), you might also want to factor in the weather forecast or a local event happening nearby (exogenous variable). These additional insights help make your sales prediction much more reliable and nuanced, just like SARIMAX improves forecasting by considering relevant external influences.
--
Key Concepts
Examples
Memory Aids
Interactive tools to help you remember key concepts
Stories
Memory Tools
Flash Cards
Glossary
SARIMA
Seasonal Autoregressive Integrated Moving Average, a time series model that captures seasonality in data.
SARIMAX
An extension of SARIMA that includes exogenous variables to improve forecasting accuracy.
Seasonality
Periodic fluctuations in data over fixed intervals, such as changes in sales during different seasons.
Exogenous Variables
External factors that can influence the dependent variable in the model.