Virginia Barnett
This week, I learned about time-series decomposition. Time-series decomposition models are one of the best forms of forecasting for businesses, because not only do they provide excellent forecasts, but, according to the text, they are consistent with the way managers tend to look at data and often helps them get a better handle on data movements by providing concrete measurements for factors that are otherwise not quantified (Keating & Wilson, 2022). The main method of time-series decomposition discussed this week was the classical time-series decomposition, which uses the ratio-to-moving average technique.
Furthermore, I learned how to go about applying this model. I found that the first step to using the model is to remove any short-term fluctuations from the data so that the long-term trends and cycles are easier to identify. These fluctuations can be seasonal patterns and irregular variations (Dagum, 2020). To remove these variations, one must calculate a moving average for the series containing the same number of periods as in the seasonality that needs identifying. This moving average will a represent a typical level of Y for the year that is centered on [the] moving average (Keating & Wilson, 2022). Furthermore, this centered moving average then represents the deseasonalized data. So then, when by comparing the actual data with the deseasonalized data, one will be able to measure the degree of seasonality.
To find the long-term trend, one will be able to use this now deseasonalized data to estimate a simple linear equation, and then from that form a centered moving-average trend. Additionally, one may also find the cyclical component using the deseasonalized data. The cyclical component is the extended wavelike movement about the long-term trend (Keating & Wilson,, 2022). The cyclical component is measured by a cycle factor, which is the ratio of the centered moving average to the centered moving-average trend (Keating & Wilson, 2022). One of the major benefits of finding this factor is that it enables forecaster to expect when the next turning point in the current cycle will be. One real-world example where this type of model is used include when study where decomposition is used to find seasonal trends in revenue development in a selected e-commerce segment based on the assessment of the applicability of the Facebook Prophet forecasting tool (Navratil & Kolkova, 2019). They do this by decomposing the data with a subsequent two-year forecast. Furthermore, what I have learned from this class is that while it is amazing and beneficial to be advancing in technology and in forecasting abilities, we should do that with the intent of keeping the overarching goal of giving God the glory. This is as 1 Corinthians 10:31 says, Whether you eat or drink or whatever you do, do it all for the glory of God (ESV 2007).
My question then is, after learning about all these different types of forecasting models so far, which one might be the best suited to a small business?
References
Dagum, E. B. (2020). Time series modeling and decomposition. Statistica, 70(4), 433-457. https://doi-org.ezproxy.liberty.edu/10.6092/issn.1973-2201/3597
Navratil, M., & Kolkova, A. (2019). Decomposition and Forecasting Time Series in the Business Economy Using Prophet Forecasting Model. Central European Business Review, 8(4), 26-39. https://doi.org/10.18267/j.cebr.221
Keating, B., & Wilson, J.H. (2022). Forecasting and Predictive Analytics (7). John Galt Solutions, Inc.
ESV Bible. (2007). Hendrickson Publishers Inc.
Jared Hoober
ThursdaySep 29 at 11:59pm
The first takeaway I had from this chapter was how it is crucial to first deseasonalize the data before creating the time-series composition model. Deseasonalizing the data to remove the short-term fluctuations allows for the long-term trend and cycle components to become clearer. The best way to do this is to use moving averages to take the seasonality out of the data. When forecasting data, it will most likely need to be deseasonalized in order to get truly accurate results. One example of a time-series decomposition is a model that was made to predict rainfall in the Wujiang River basin in China. This model had to take into account the obvious seasonality of the data because precipitation varies between the seasons (Wang et. al., 2022). Deseasonalizing data not only allows the forecaster to identify the trends and patterns in the data, but also provides a helpful overall measurement of seasonality using indexes. One thing to note is that sometimes certain period indexes cannot be calculated so proper planning and identification needs to be done first. As forecasters, it can be challenging to not constantly be concerned about what is coming next because that is literally our jobs. However, as Christians we are commanded to not worry about what may happen tomorrow. Matthew 6:25 says, Therefore I tell you, do not worry about your life, what you will eat or drink; or about your body, what you will wear. Is not life more than food, and the body more than clothes? (New International Version, 1984, Matthew 6:25). This is an awesome reminder that whatever happens we do not need to be concerned with what tomorrow will hold because we can trust that God will take care of it.
Another takeaway I had from this chapter was how the cyclical component of the model is really important especially in the business sense. This component of the model is the most difficult to analyze and properly forecast in the period. Every business has a business cycle, and it must be incorporated into the model. Luckily, with the advancement in technology and AI it is becoming increasingly easier to discover cyclical patterns in data (Gozuyilmaz et al., 2021). Forecasters have a lot to be excited about in the coming years with much more useful tools right at their fingertips to create and analyze models.
Question: What are some of the advantages and disadvantages of a time-series decomposition and how does this model compare to models from prior chapters?
Gozuyilmaz, S., & Kundakcioglu, O. E. (2021). Mathematical Optimization for Time Series Decomposition. OR Spectrum, 43(3), 733758.
The Holy Bible, New International Version. (1984). Grand Rapids: Zondervan Publishing House
Wang, Y., Liu, J., Li, R., Suo, X., & Lu, E. (2022). MEEMD DecompositionPredictionReconstruction Model of Precipitation Time Series. Sensors (14248220), 22(17), 6415. https://doi.org/10.3390/s22176415
The student
must then post 2 replies of at least 200-300 words by 11:59 p.m. (ET) on Monday of the assigned
Module: Week. For each thread, students must support their assertions with at least 2 scholarly
citations other than the textbook and at least 1 biblical integration in APA format. Each reply
must incorporate at least 2 scholarly citations other than the textbook and at least 1 biblical
integration in APA format. Any sources cited must have been published within the last five
years. Acceptable sources include the textbook, the Bible, etc.