You are on page 1of 3

Moving Average:

State the conditions under which the Moving Average method can be recommended
for trend analysis. How will you determine the period of the moving average?

The moving average (MA) method is a widely used statistical method for trend
analysis. It can be recommended in the following conditions:

Stable trend: Moving averages effectively smooth out fluctuations in time series data
and identify underlying trends. This is most useful when the trend is relatively stable
over time. If the data exhibits significant seasonality, cyclical patterns, or irregular
fluctuations, other methods such as smoothing or exponential decay may be more
appropriate. Landline phone data. Stationary data is data whose statistical properties
such as mean and variance do not change over time. Moving averages work best with
fixed data. Non-stationary data can be converted to stationary data by applying
differential or other methods before applying a moving average.

Interval of consecutive data collection. Data must be collected periodically (e.g. daily,
monthly, yearly). Unevenly distributed data may require additional preprocessing.

Enough data points. To calculate meaningful moving averages, you must have enough
historical data points. The number of data points needed depends on the frequency of
the underlying trend. Longer historical data series often yield more information.
Apparent periodic behavior: Moving averages are used to identify and quantify trends
that occur over a certain period of time. Therefore, you need to have a clear
understanding of the periodic behavior of the trend you want to analyze. For example,
if you are interested in yearly trends, use a yearly moving average.

Avoid lag: Moving averages inherently introduce lag into the data. The magnitude of
this lag depends on the choice of moving average period. You should be aware of this
delay and consider whether it is acceptable for your analysis. Shorter moving
averages have less lag but can be noisier, while longer moving averages offer
smoother but higher latency.

To determine the moving average period, you can:

Consider the nature of the data: think about the cyclicality of the underlying trend. For
example, if you are analyzing monthly sales data and suspect a year-over-year trend,
you might start with a 12-month moving average.

Experiment: Try different moving average periods and see how they affect your
smoothed data. Shorter time periods (e.g., 3, 6, or 12 months) provide greater
sensitivity to recent changes but can be disruptive. Longer time periods (e.g. 24 or 36
months) yield smoother trends but with longer lags.

Use domain knowledge: If you have knowledge of the domain or historical context of
the data, this can help you choose a moving average period. For example, you may
know that your industry has a six-month seasonal pattern, which suggests a six-month
moving average.

Evaluate results: Evaluate the effectiveness of different moving average periods in


identifying the underlying trend. You can use statistical measurements or visual
inspection to determine which time periods provide the most meaningful information.

Consider the trade-off: Remember that there is often a trade-off between


responsiveness to recent changes and smoothness. Choose a period that provides
appropriate balances for analysis purposes.

Ultimately, the choice of moving average period must be tailored to the specific data,
the nature of the trend being analyzed, and the goals of the trend analysis.

(a) Calculate the 4-year moving average of the following data related to the sales in a
department store.

Average
Year Sales Sales (t-4)
2000 960
2001 976
2002 974
2003 996 976.5
2004 1024 992.5
2005 1040 1008.5
2006 1688 1187
2007 1128 1220
2008 1144 1250
2009 1120 1270
2010 1140 1133
2011 1168 1143
2012 1196 1156
2013 1212 1179
2014 1200 1194
2015 1180 1197

(b) Plot a time series graph for the above data.


2000-2015 Sales
1800
1600
1400
1200
Sales

1000
800
600
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
Year

Sales Average Sales (t-4)

(c) Plot the corresponding trend line using the method of moving average.

2000-2015 Sales
1800
1600
1400
1200
Sales

f(x) = 14.5521978021978 x + 1001.13186813187


1000
800
600
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
Year

Sales Average Sales (t-4)


Linear (Average Sales (t-4))

You might also like