SECTION 4 TME SERIES ANAYSIS AND

CONTROL CHARTS
Structure Page Nos.
4.0 Introduction 60
4.1 Objective 60
4.2 Time Series and Forecasting 60
4.3 Control Charts 66
4.4 Summary 69
4.5 Answers to Check Your Progress 69
4.6 Further Readings 69

4.0 INTRODUCTION
In the previous section, you were introduced to regression analysis, where it was
desired to fit a linear (in case of linear regression) or a curve between the dependent
and the independent variables. A typical decision making through use of regression
analysis may involve plan for future growth of infrastructure of a city based on the
growth pattern in population. Thus, prediction primarily focuses on estimating a
future value based on the presented study data.
This section introduces the concept of Time series, which you learnt in BCS040/Block
3, and also, how they can be used for forecasting. The Unit begins with an
introduction to time series and then shows how you can use a worksheet to forecast
using time series data.
Further, the unit introduces the control charts that are used in determining the
statistical quality control. You must go through the following two units of
BCS040/Block 3 before performing various tasks/steps/activities listed in this section.
Unit 10: Forecasting and Time Series Analysis
Unit 11: Statistical Quality Control

4.1 OBJECTIVES

After performing the activities of this section, you should be able to:
 define time series data;
 use a spreadsheet to create time series data;
 use spreadsheet to fit the linear trends using the method of least squares;
 use spreadsheet for forecasting models.

4.2 TIME SERIES AND FORCASTING

In many situations, data may be collected over a period of time. This time period may
be in terms of days, for example, for shopping analysis you may be interested in
recording sales on Mondays, Tuesdays, etc.; a week when you want to plan weekly, a
month, a year and so on. Such periodic data is called times series data. Block
3/Unit10/ Page44/ Table 3 which is reproduced in the Figure 1 shows the time series
data of crop yields. This data is recorded for a time period of a year over 41 years.
This data was collected by Government of AP for studying the changes in the
cropping pattern. Such data may be used for first predicting a trend. In this example, it
is expected to predict the future economic needs of the Agricultural sector.
60
 Time Series Analysis
and Control Charts

Figure 1: Long Term Trends for Crop yield

The figure shows the crop yield data and the long term trend for two crops Rice and
Sugarcane over a period of 41 years. The Figure also shows the trend line of the two
crop yields. To draw the graph, we have selected a linear Scatter plot. In the Select
Data… option you can name the Series viz. “Rice” and “Sugarcane. For checking the
details of series in a spreadsheet package, right click on the chart, and select Select
Data… option. You will see the following window:

You may edit any series, for example, if you select Rice Data Series and click on Edit
Button, you will see the following Range for the series:

61
Statistical
Techniques Lab Thus, you can modify the series as per your requirements. Now select any series on
the chart and press right mouse button, the following Menu will be displayed:

Using this Menu, you can change series or chart type, select data, add data labels and
format data series. You may explore these options on your own. One option which
requires a mention due to its usefulness here is Add Trendline… On selecting this
option, you will get the following Menu:

Please note that for Figure 1, you can select Trendline option as Linear. You can also
change the Line Color and Line Style using options shown after arrow.

Figure 1(a): Trendline Options

62
Now observe the data series in Figure 1. There are fluctuations in the time series data. Time Series Analysis
You can attribute these fluctuations on various factors that may include availability and Control Charts
and quality of fertilizer, weather conditions etc. You may refer to page 44/Unit 10 for
a complete discussion. We briefly present below the following:
Long term trends: From the Figure 1, the long term trend that can be drawn is that
average yield of Rice crop per hectare has increased. On the other hand, the yield of
Sugarcane per hectare has steadily dropped. These trends can be further analysed to
ascertain the reasons of such the declining trends.
Another long term trend about rainfall is shown in Figure 2. You can see that average
rainfall over the period of 41 years is almost unchanged.

Figure 2: Trendline for Average Rainfall

Seasonal Variations: As a task, to start with, create the data for the seasonal variations
given in the BCS040: Block 3/Unit 10/Page 46 and plot the time series in two possible
ways. (i) Plot the figure akin to Fig.5/Unit 10 for all the six years, instead of three
shown, with only four quarters along the x-axis. (ii) Plot the time series with years
along the x-axis, showing quarters within each of the years. Note the variations that
you observe in the curve drawn by the spreadsheet. Discuss with other students or
counselors the “relevance” of seasonal variation in data besides commonly
encountered examples in real life.

Cyclic variations: Using the data on Sugarcane Area, the cyclic variations are plotted
in Figure 3 using spreadsheet package. You can observe the cyclic pattern in the plot.
Did you understand the distinction between “a season” and “a cycle”, which is
discussed in section 10.3.3/page 48/Unit 10? Discuss the key points from there with
other students or counselors.

63
Statistical
Techniques Lab

Figure 3: Cyclic variations

Forecasting Models: Please go through various forecasting models from Section
10.4/page 49/MCS040/Block3/Unit 10, viz., additive and multiplicative models. You
may observe that Trendline actually is a forecasting curve line. In Figure 1 and Figure
2, you have used a linear fit. You can revise from Unit 9 or MCS040/Block3/Unit
10.5.1 the method of least squares method. Further, you can select the method of
moving averages (MCS040/Block3/Unit 10.5.2) by simply selecting the Trendline
Options as shown in Figure 1(a) or you can fit the Polynomial of second degree using
least square method, by simple selecting Polynomial Options and keeping the order as
2 (see Figure 4).

Figure 4: The Polynomial Trendline of Order 2

64
You can also select the Moving Average Trendline option from the same window. Time Series Analysis
Another alternative approach for calculating moving averages is that you can use Data and Control Charts
 Data Analysis  Moving Average option. On selection of this option, you will be
presented the following Moving Average Window :

Select the input range, interval and output range. If you want a chart of moving
averages then you can select Chart Output option box. Figure 5 shows moving
averages with various intervals using repeated application of this command.

Figure 5: Finding Moving Averages

Thus, you can perform the long term trend analysis using spreadsheet packages.
 Check Your Progress 1
1) Develop all the spreadsheets as shown in Figures 1 to Figure 5.
.........................................................................................................................................
.........................................................................................................................................
.........................................................................................................................................
.........................................................................................................................................
.........................................................................................................................................

65
Statistical  Lab Sessions 7
Techniques Lab
1) Workout all the Examples given in the Unit 10 using spreadsheet package.
2) Workout the Exercises of BCS040/Blockl2/Unit 10 using spreadsheet wherever
required, or generate sample data wherever possible using spreadsheet functions
of random number generation. Alternatively, look for other examples from
references with sample data.

4.3 CONTROL CHARTS

This section discusses about the use of a spreadsheet package in constructing a control
chart. You should go through BCS-040 Block 3/ Unit 11 before going through this
section. Let us first revise some of the terms used in that Block.

Quality and Quality characteristics: The term Quality may broadly be defined as
conformance of a product to certain standards. Quality depends on certain
characteristics related to the product. When you design a product, certain
specifications or levels of tolerances are established on all important quality
characteristics of the product. These characteristics are called the quality. For
example, in case of a ball pen, the specifications on refill length may be that it should
lie between 9.90 crns and 10.10 crns.

Controlling a Process: The control operations of a process has to be such that the
quality characteristics of output product are maintained at the desired levels.

Statistical Process Control: It is a methodology that collects and analyses data on

quality characteristics periodically from the process, to take appropriate actions in
respect of quality monitoring of a process. Control charts are a most widely used
technique for Statistical Process Control.

Variations and Stable/unstable process: A certain amount of variation in quality

characteristics (for example, length of refill) is unavoidable despite stable conditions
(for example, same machine settings, almost identical material quality and experience
of operators). The small changes in quality characteristics may be attributed to small
conditional changes. These small causes are inevitable and are called chance causes.
The variation caused by such changes is called chance cause variation. However,
some major chances, such as change in the machine settings, major drop in the quality
of raw material etc., disturb the stable process. Such variations must be identified and
corrected. Since, you can attribute change to one or more particular reasons, they are
called assignable causes. Such causes may make a process unstable.

Control Chart Construction steps:

 Collect sample periodically

 Compute quality characteristics for the collected sample
 Plot sample number ( -axis) vs quality characteristic ( -axis)
 Check if the value of quality characteristics is between the upper and lower
control limits.
 In general, the quality characteristics follow normal distribution. Therefore, the
upper and lower limits may be determined by using “Three Sigma Limits” as:
lower limit and upper limit .
 If quality characteristics are a variable, then control in terms of both the mean and
the variability is sought.

66
Control chart can tell us when process has become unstable, but not that why did it Time Series Analysis
happen? and Control Charts

Problem 1: (Reference MCS040/Block 3/Unit 11/Page 83), Data for the length of
refills and control chart constants (only for sample size ( ) are shown in Figure 6.
This data is drawn from the Table 1 and Table 2 of the said Unit. You may now
estimate and from the refill length data and calculate the control limits for and
charts.

The said Unit shows the calculations that are performed with the help of spreadsheet
and shown in Figure 6.

Figure 6: Calculation of µ and σ for refill length

For creating Figure 6, you can use your own formulas.

Steps

 You first create the sample data.

 The mean of the sample subgroup at a particular time is found in cell J3
using the formula and similarly
 the Range for this subgroup is calculated at cell K3 using the formula
. For .
 On the basis of these mean and range values and is calculated in the
cell O5 using the formula =SUM(J3:J22)/O4 and cell O6 using the
formula =SUM(K3:K22)/O4 respectively.
 On the basis of these values and control chart constants CL, LCL and
UCL values are calculated for the and chart. Write the formula for
these calculations and compare the values with the values calculated in
the Unit.

67
Statistical Further,
Techniques Lab
 on the basis of these calculated values, you can create data as shown in
Figure 7.
 This data is used to draw the control chart.
 You may use Scatter Plot to draw the charts as shown in Figure 7.
 You may keep the x-axis same for all the Ranges that are being plotted
as Y axis.

Please note that in Figure 7, that the 12th point on the R-chart crosses the UCL line,
which is indicative of “process out of control” situation with respect to variability (as
this problem has been detected in R-chart). Consequently, we conclude that during
this time period the process may not be stable.

Figure 7: Control Charts (x-chart and R-chart) for the refill length data

Now you should try the following lab session:

 Check Your Progress 2

1) Develop the spreadsheets as shown in Figures 6 and Figure 7.

................................................................................................................................................
................................................................................................................................................
................................................................................................................................................
................................................................................................................................................
................................................................................................................................................

68
 Lab Sessions 8 Time Series Analysis
and Control Charts
1) Perform all the examples given in the Unit 11 using spreadsheet package
including example given in section 11.3.5 on control charts for attributes.

2) Perform all Exercises of BCS040/Block3/Unit 11 that you can perform using

spreadsheet.

3) Write simple C functions to calculate various values required for drawing control
charts.

4.4 SUMMARY
This section was an attempt to provide you details on some of the elementary methods
of Unit 10 based on which you can perform time series analysis and forecasting using
the spreadsheet. Time series analysis helped in performing analysis of data in terms of
the components and the models discussed in the said Unit 10. The example
demonstrated the use of spreadsheet package functions in evaluating trend.

This section also covered computational aspects relating to control charts in statistics
using spreadsheet package. The purpose of these charts was to determine the process
under statistical control.

4.5 ANSWERS TO CHECK YOUR PROGRESS

Check Your Progress 1:

Please use any spreadsheet package and enter all the data as shown in Figure 1
to Figure 5. You may download the file containing all these figures from the
website www.ignou.ac.in from BCA page under MCSL044. Find out what
formulas have been entered. Also find the errors in the entry of formula.

Check Your Progress 2:

Please use any spreadsheet package and enter all the data as shown in Figure 6
and Figure 7. You may download the file containing all these figures from the
website www.ignou.ac.in from BCA page under MCSL044. Find out what
formulas have been entered. Also find the errors in the entry of formula.

4.6 FURTHER READINGS

1. BCS-040: Statistical Techniques, IGNOU Material.

Web link:
 www.wikipedia.org

69