Trend Analysis

Overview for Trend Analysis

Learn more about Minitab
Use Trend Analysis to fit a general trend model to time series data and to provide forecasts. You can choose
between the linear, quadratic, exponential growth or decay, and S-curve trend models. Use this procedure to fit
a trend when your data have a very consistent trend and no seasonality.
For example, a marketing analyst uses a trend analysis to predict the sales of tennis racquets for the next 3
months.
Where to find this analysis
• Mac: Statistics > Time Series > Trend Analysis
• PC: STATISTICS > Forecast > Trend Analysis
When to use an alternate analysis
If your data do not have a trend and do not have a seasonal component, use Moving Average or Single
Exponential Smoothing.

Data considerations for Trend Analysis

Learn more about Minitab
To ensure that your results are valid, consider the following guidelines when you collect data, perform the
analysis, and interpret your results.
Record data in chronological order
Time series data are collected at regular intervals and are recorded in time order. You should record the data in
the worksheet in the same order that you collect it. If the data are not in chronological order, you cannot assess
time-related patterns in the data. However, you can still use Scatterplot to investigate the relationship between
a pair of continuous variables.
Collect enough data to assess trends or patterns
Collect enough data so that you can fully assess trends or patterns in the data. For example, you need enough
data to be sure that any pattern you observe is a long-term pattern and not just a short-term anomaly.
Collect data at appropriate time intervals
Choose the time interval based on the patterns that you want to detect. For example, to look for month-to-month
patterns in a process, collect data at the same time each month. If you collect data each week, then the monthly
pattern may be lost in the noise of the weekly data. If you collect data each quarter, the monthly pattern may be
lost when it is averaged out in each quarter.
If you are looking only for general trends or shifts in the data over time, and not for patterns associated with a
specific time interval, the length of the interval is less important.
Your data should have a trend with no seasonal component
If your data do not have a trend and do not have a seasonal component, use Moving Average or Single
Exponential Smoothing.
The trend should follow a consistent shape without shifts or reversals
If your data have cyclical movements, shifts, or reversals in the trend, use Double Exponential
Smoothing because it uses a dynamic trend component that works well with changes in the trend.

Example of Trend Analysis

Learn more about Minitab
A marketing analyst wants to assess trends in tennis racquet sales. The analyst collects sales data from the
previous five years to predict the sales of the product for the next 3 months.
1. Open the sample data, TennisRacquetSales.MTW.
2. Open the Trend Analysis dialog box.
• Mac: Statistics > Time Series > Trend Analysis
• PC: STATISTICS > Forecast > Trend Analysis
3. In Y variable, enter Racquets.
4. Under Type of model, select Quadratic.
5. On the Forecast tab, select Number of forecasts to generate and enter 3.
6. Click OK.
Interpret the results
The trend analysis plot shows a general upward trend. There is curvature and the model appears to fit the data
well, which indicates a good fit. The analyst can be confident in the forecasts for the next 3 months.

Method

Model type Quadratic

Rows used 60
Fitted Trend Equation

Yt = 324.352 + 0.491563 × t + 0.0469355 × t2

Accuracy Measures
MAPE 2.47
MAD 9.46
MSD 135.70
Forecasts

Period Forecast
61 529.0
62 535.2
63 541.6
Perform the analysis
Enter your data for Trend Analysis (Data tab)
Learn more about Minitab
On the Data tab of the Trend Analysis dialog box, specify the data for the analysis, specify the time scale, and
select the type of model.
In This Topic
• Enter your data
• Time scale labels
• Type of model
Enter your data
In Y variable, enter a column of numeric data that were collected at regular intervals and recorded in time order.
If your data are in multiple columns (for example, you have data for each year in a separate column), you must
stack the data into a single column.
In this worksheet, Sales contains the number of computers that are sold each month.

C1
Sales
195000
213330
208005
249000
237040
Time scale labels
(Optional) Enter a column to label the x-axis with values, such as dates. If you don't enter a column, Minitab
labels each time period with an integer starting at 1.
Type of model
Under Type of model, specify the type of model that matches the trend in your data. To choose between these
four models:
• Graph the data using a time series plot. Then, compare your plot to the following figures to determine
the correct type of model.
• You can fit the four models and compare the measures of accuracy (MAPE, MAD, and MSD). Choose
the model with the smallest accuracy measures.
Linear
The data fits a line, which indicates that the rate of change is uniform over time. The model is Yt = β0 + (β1 * t)
+ et. In this model, β1 represents the average change from one period to the next.

Quadratic
The data have a curvature, which indicates that the rate of change varies over time. The model is Yt = β0+ β1 *
t + (β2* t2) + et.

Exponential growth
The data have a steep curvature, which indicates that the rate of change varies more quickly over time. For
example, a savings account might exhibit exponential growth. The model is Yt = β0 + (β1t) + et.

S-curve (Pearl-Reed logistic)

The data has an S-shape, which indicates that the direction of the change varies over time. The model is Y t =
(10a) / (β0 + β1 * β2t).
Generate forecasts for Trend Analysis (Forecast tab)
Learn more about Minitab
On the Forecast tab of the Trend Analysis dialog box, complete the following steps to generate forecasts for
your time series.
1. Select Number of forecasts to generate and enter the number of consecutive time periods that you
want forecasts for.
2. (Optional) Select Generate from origin and enter the row number for the first forecast. Minitab uses
only the data up to that row number for the forecasts. The forecast values differ from the trend line
because Minitab uses all of the data to calculate the trend line values.

Select the graphs for Trend Analysis (Graphs tab)

Learn more about Minitab
On the Graphs tab of the Trend Analysis dialog box, select the graphs to include in your output.
Residual plots
Select to have Minitab display all four plots.
Residuals versus fits
Display the residuals versus the fitted values. Use the residuals versus fits plot to determine whether the residuals
are unbiased and have a constant variance.
Residuals versus order
Display the residuals versus the order of the data. The row number for each data point is shown on the x-axis.
Use the residuals versus order plot to see how accurate the fitted values are compared to the observed values
during the observation period.
Normal probability plot of residuals
Display a normal probability plot of the residuals. Use the normal plot of residuals to determine whether the
residuals are normally distributed. However, the residuals being normally distributed is not an assumption that
has to be met for this analysis.
Histogram of residuals
Display the shape and spread of the residuals. Use the histogram of residuals to determine whether the data are
skewed or whether outliers exist in the data.
Residuals versus variables
Enter one or more variables to plot versus the residuals. Use the residuals versus the variables plot to assess
whether a variable that is not in the analysis systematically affects the response.
Interpret the results
Interpret the key results for Trend Analysis
Learn more about Minitab
Complete the following steps to interpret a trend analysis. Key output includes the fitted trend equation, the
accuracy measures, and the forecasts.
In This Topic
• Step 1: Determine whether the model fits your data
• Step 2: Compare the fit of your model to other models
• Step 3: Determine whether the forecasts are accurate
Step 1: Determine whether the model fits your data
Examine the trend analysis plot to determine whether your model fits your data. If the fits closely follow the
actual data, the model fits your data. Ideally, the data points should fall randomly around the fitted line.
• If the model fits the data, you can perform double exponential smoothing and compare the two models.
• If the model does not does fit the data, perform the analysis again and select a different type of model.
If you fit a linear model and see curvature in the data, select the quadratic, exponential, or S-curve
model. If none of the models fit your data, use a different time series analysis. For more information,
go to Which time series analysis should I use?.

On this trend analysis plot, the fits closely follow the data, which indicates that the model fits the data.
Step 2: Compare the fit of your model to other models
Use the accuracy measures (MAPE, MAD, and MSD) to compare the fit of your model to other time series
models. These statistics are not very informative by themselves, but you can use them to compare the fits
obtained by using different methods. For all 3 statistics, smaller values usually indicate a better fitting model.
If a single model does not have the lowest values for all 3 statistics, MAPE is usually the preferred measurement.
NOTE
The accuracy measures provide an indication of the accuracy you might expect when you forecast out 1 period
from the end of the data. Therefore, they do not indicate the accuracy of forecasting out more than 1 period. If
you're using the model for forecasting, you shouldn't base your decision solely on accuracy measures. You
should also examine the fit of the model to ensure that the forecasts and the model follow the data closely,
especially at the end of the series.
Model 1 Model 2
Accuracy Measures Accuracy Measures

MAPE 8.1976 MAPE 6.9551

MAD 3.6215 MAD 2.7506
MSD 22.3936 MSD 11.2702

Key Results: MAPE, MAD, MSD

In these results, all three numbers are lower for the 2nd model compared to the 1st model. Therefore, the
2nd model provides the better fit.
Step 3: Determine whether the forecasts are accurate
Examine the end of the trend analysis plot and the forecasts to determine whether the forecasts are likely to be
accurate. The fits should follow the data closely, especially at the end of the series. If the fits start to shift away
from the data at the end of the series, the underlying trend may be changing. If the trend is changing, the model
might not generate accurate forecasts. In this case, collect more data to determine whether the trend over a
longer period of time is less consistent.
Even if your forecasts appear to be accurate, be cautious about forecasts that are more than 3 periods in the
future. Trends observed over a short span of data could be part of a larger cycle and may not persist into the
future. Because trends can be volatile, you should usually only forecast 2 or 3 periods into the future.

Forecasts

Period Forecast
61 529.0
62 535.2
63 541.6

On this trend analysis plot, the fits closely follow the data, especially at the end of the series. You can conclude
that the forecasts for the next 3 months are accurate.
Interpret all statistics and graphs for trend analysis
Learn more about Minitab
Find definitions and interpretation guidance for every statistic and graph that is provided with trend analysis.
In This Topic
• Rows used • Forecast
• Rows not used • Trend analysis plot
• Fitted Trend Equation • Normal probability plot of the residuals
• Curve parameters • Residuals versus fits
• MAPE • Histogram of the residuals
• MAD • Residuals versus order
• MSD • Residuals versus variables
• Period
Rows used
The number of observations in the time series.
Rows not used
The number of missing values in the time series.
Fitted Trend Equation
Use the fitted trend equation to describe how the variable that you measure changes over time. The fitted trend
equation is an algebraic representation of the trend line. The form of the fitted trend equation depends on the
type of model that you selected.

Type of model Equation

Linear Yt = b0 + (b1 * t)

Quadratic Yt = b0+ b1 * t + (b2* t2)

Exponential growth Yt = b0 + (b1t)

S-Curve (Pearl-Reed logisitic) Yt = (10a) / (b0 + b1 * b2t).

In the fitted trend equation, the letters represent the following:
• yt is the variable
• b0 is the constant
• b1 and b2 are the coefficients
• t is the value of the time unit
Curve parameters
When you choose the S-curve model, Minitab displays the parameters for the model. The model is Yt = (10a) /
(β0 + β1 * β2t).
Intercept
The value of the model at time = 0. The intercept equals 1/β0 + β1.
Asymptote
The value that the model approaches as time increases to infinity. The asymptote equals 1/β0.
Asymptotic Rate
The rate at which the model approaches the asymptote. Models with lower values approach the asymptote faster.
The asymptotic rate equals β2.
MAPE
The mean absolute percent error (MAPE) expresses accuracy as a percentage of the error. Because the MAPE
is a percentage, it can be easier to understand than the other accuracy measure statistics. For example, if the
MAPE is 5, on average, the forecast is off by 5%.
However, sometimes you may see a very large value of MAPE even though the model appears to fit the data
well. Examine the plot to see if any data values are close to 0. Because MAPE divides the absolute error by the
actual data, values close to 0 can greatly inflate the MAPE.
Interpretation
Use to compare the fits of different time series models. Smaller values indicate a better fit. If a single model
does not have the lowest values for all 3 accuracy measures, MAPE is usually the preferred measurement.
The accuracy measures are based on one-period-ahead residuals. At each point in time, the model is used to
predict the Y value for the next period in time. The difference between the predicted values (fits) and the actual
Y are the one-period-ahead residuals. Because of this, the accuracy measures provide an indication of the
accuracy you might expect when you forecast out 1 period from the end of the data. Therefore, they do not
indicate the accuracy of forecasting out more than 1 period. If you're using the model for forecasting, you
shouldn't base your decision solely on accuracy measures. You should also examine the fit of the model to
ensure that the forecasts and the model follow the data closely, especially at the end of the series.
MAD
The mean absolute deviation (MAD) expresses accuracy in the same units as the data, which helps conceptualize
the amount of error. Outliers have less of an effect on MAD than on MSD.
Interpretation
Use to compare the fits of different time series models. Smaller values indicate a better fit.
The accuracy measures are based on one-period-ahead residuals. At each point in time, the model is used to
predict the Y value for the next period in time. The difference between the predicted values (fits) and the actual
Y are the one-period-ahead residuals. Because of this, the accuracy measures provide an indication of the
accuracy you might expect when you forecast out 1 period from the end of the data. Therefore, they do not
indicate the accuracy of forecasting out more than 1 period. If you're using the model for forecasting, you
shouldn't base your decision solely on accuracy measures. You should also examine the fit of the model to
ensure that the forecasts and the model follow the data closely, especially at the end of the series.
MSD
The mean square deviation (MSD) measures the accuracy of the fitted time series values. Outliers have a greater
effect on MSD than on MAD.
Interpretation
Use to compare the fits of different time series models. Smaller values indicate a better fit.
The accuracy measures are based on one-period-ahead residuals. At each point in time, the model is used to
predict the Y value for the next period in time. The difference between the predicted values (fits) and the actual
Y are the one-period-ahead residuals. Because of this, the accuracy measures provide an indication of the
accuracy you might expect when you forecast out 1 period from the end of the data. Therefore, they do not
indicate the accuracy of forecasting out more than 1 period. If you're using the model for forecasting, you
shouldn't base your decision solely on accuracy measures. You should also examine the fit of the model to
ensure that the forecasts and the model follow the data closely, especially at the end of the series.
Period
Minitab displays the period when you generate forecasts. The period is the time unit of the forecast. By default,
the forecasts start at the end of the data.
Forecast
The forecasts are the fitted values that are obtained from the time series model. Minitab displays the number of
forecasts that you specify. The forecasts begin either at the end of the data or at the point of origin that you
specify. Minitab uses the data before the point of origin to calculate the coefficients of the fitted trend equation.
If you specify a point of origin, Minitab uses only the data up to that row number for forecasts.
Interpretation
Use forecasts to predict a variable for a specified period of time. For example, a warehouse manager can model
how much product to order for the next 3 months based on the previous 60 months of orders.
Examine the end of the trend analysis plot and the forecasts to determine whether the forecasts are likely to be
accurate. The fits should follow the data closely, especially at the end of the series. If the fits start to shift away
from the data at the end of the series, the underlying trend may be changing. If the trend is changing, the model
might not generate accurate forecasts. In this case, collect more data to determine whether the trend over a
longer period of time is less consistent.
Even if your forecasts appear to be accurate, be cautious about forecasts that are more than 3 periods in the
future. Trends observed over a short span of data could be part of a larger cycle and may not persist into the
future. Because trends can be volatile, you should usually only forecast 2 or 3 periods into the future.
Trend analysis plot
The trend analysis plot displays the observations versus time. The plot includes the fits calculated from the fitted
trend equation, the forecasts, and the accuracy measures.
Interpretation
Examine the trend analysis plot to determine whether your model fits your data. If the fits closely follow the
actual data, the model fits your data. Ideally, the data points should fall randomly around the fitted line.
• If the model fits the data, you can perform double exponential smoothing and compare the two models.
• If the model does not does fit the data, perform the analysis again and select a different type of model.
If you fit a linear model and see curvature in the data, select the quadratic, exponential, or S-curve
model. If none of the models fit your data, use a different time series analysis. For more information,
go to Which time series analysis should I use?.
On this trend analysis plot, the fits closely follow the data, which indicates that the model fits the data.
Normal probability plot of the residuals
The normal plot of the residuals displays the residuals versus their expected values when the distribution is
normal.
Interpretation
Use the normal plot of the residuals to determine whether the residuals are normally distributed. However, this
analysis does not require normally distributed residuals.
If the residuals are normally distributed, the normal probability plot of the residuals should approximately follow
a straight line. The following patterns imply that the residuals are not normally distributed.

S-curve implies a distribution with long tails.

Inverted S-curve implies a distribution with short tails.

Downward curve implies a right-skewed distribution.

A few points lying away from the line implies a distribution with outliers.
Residuals versus fits
The residuals versus fits plot displays the residuals on the y-axis and the fitted values on the x-axis.
Interpretation
Use the residuals versus fits plot to determine whether the residuals are unbiased and have a constant variance.
Ideally, the points should fall randomly on both sides of 0, with no recognizable patterns in the points.
The patterns in the following table may indicate that the residuals are biased and have a nonconstant variance.

What the pattern may

Pattern indicate

Fanning or uneven spreading of residuals across fitted Nonconstant variance

values

Curvilinear A missing higher-order term

A point that is far away from zero An outlier

If you see nonconstant variance or patterns in the residuals, your forecasts may not be accurate.
Histogram of the residuals
The histogram of the residuals shows the distribution of the residuals for all observations. lf the model fits the
data well, the residuals should be random with a mean of 0. So the histogram should be approximately
symmetric around 0.
Residuals versus order
The residuals versus order plot displays the residuals in the order that the data were collected.
Interpretation
Use the residuals versus order plot to determine how accurate the fits are compared to the observed values
during the observation period. Patterns in the points may indicate that model does not fit the data. Ideally, the
residuals on the plot should fall randomly around the center line.
The following patterns may indicate that the model does not fit the data.

Pattern What the pattern may indicate

A consistent long-term trend The model does not fit the data

A short-term trend A shift or a change in pattern

A point that is far away from the other points An outlier

A sudden shift in the points The underlying pattern for the data has changed
The following examples show patterns that may indicate that the model does not fit the data.

Residuals systematically decrease as the order of the observations increases from left to right.

A sudden change in the values of the residuals occurs from low (left) to high (right).
Residuals versus variables
The residuals versus variables plot displays the residuals versus another variable.
Interpretation
Use the plot to determine whether the variable affects the response in a systematic way. If patterns are present
in the residuals, the other variables are associated with the response. You can use this information as the basis
for additional studies.

Methods and formulas

Methods and formulas for Trend Analysis
Learn more about Minitab
Select the method or formula of your choice.
In This Topic
• Linear • Forecasts
• Quadratic • MAPE
• Exponential growth • MAD
• S-curve • MSD

Linear Term Description

Formula β0 the constant
The linear trend model is: β1 average change from one period to the next
Yt = β0 + β1 t + et t value of the time unit
et the error term
Quadratic
Formula
The quadratic trend model, which can account for simple Term Description
curvature in the data, is:
β0 the constant
Yt = β0 + β1 t + β2 t + et
2
β1 and β2 the coefficients
t value of the time unit
et the error term
Exponential growth
Formula
The exponential growth trend model accounts for exponential growth or decay. For example, a savings account
might exhibit exponential growth.
Yt = β0 β1t + et Term Description
β0 the constant
β1 the coefficient
t value of the time unit
S-curve
et the error term
Formula
The data has an S-shape, which indicates that the direction of the change varies over time.
Yt = 10a / (β0 + β1 β2t )
Term Description
β0 the constant
β1 and β2 the coefficients
t value of the time unit
Forecasts
Minitab uses the trend equation to calculate the forecast for specific time values. Data before the forecast origin
are used to fit the trend.

MAPE
Mean absolute percentage error (MAPE) measures the accuracy of fitted time series values. MAPE expresses
accuracy as a percentage.
Formula Term Description
yt actual value at time t
fitted value

n number of observations
MAD
Mean absolute deviation (MAD) measures the accuracy of fitted time series values. MAD expresses accuracy
in the same units as the data, which helps conceptualize the amount of error.
Formula
Term Description
yt actual value at time t
fitted value

n number of observations
MSD
Mean squared deviation (MSD) is always computed using the same denominator, n, regardless of the model.
MSD is a more sensitive measure of an unusually large forecast error than MAD.
Formula
Term Description
yt actual value at time t
fitted value

n number of observations