Time Series Analysis and Forecasting - Delphi Technique

- Visionary Forecast
Accuracy - Product life cycle analysis

Time Series and Forecasting: Overview 2. Quantitative

a. objective, which makes use of quantitative or
PAST numerical data to make a forecast.
Historical Information b. suitable when variable involved has an available
historical quantitative relevant data.
PRESENT According to Ross:
Analysis of the Past information to uncover design Time Series (Intrinsic)
or pattern or model. -Simple Average
- Moving Average
FUTURE - Exponential Smoothing
Forecasting or predicting possible emerging event -Time Series Decomposition
in the future based on the model made at present Associative (Extrinsic)
based on the analyzed past information. -Regression
-Multiple Regression
Major Area of Forecasting -Leading Indicator
-Econometric
Demand Forecasting
Predicts timing and quantity of demand of firm’s Concept of Time Series Analysis
commodities Singh and Singhal reiterated in their paper
presentation the conceptual description of time
Economic Forecasting series by Ya Lun Chau (2014):
Predicts future business conditions with reference
to the underlying economic factors such as A time series is…
inflation, GDP,GNP, etc. - A set of data depending on time
- A series of values over a period of time
Technology Forecasting - Collection of magnitude belonging to
Predicts possible technological advancements in different time periods of some variable or
the future such latest digital devices and systems, composite of variables such as production
future software and hardware, etc. of steel, per capita income, gross
international income, price of tobacco,
Forecasting Approaches index of industrial production.
(Reid & Sanders, 2012)
Classification of Forecasting Approaches  Time is act as a device to set of common
1. Qualitative stable reference point.
a. subjective, makes use of professional prudence  In time series, time act as an independent
in making forecasts. variable to estimate dependent variables.
b. suitable if the variable involved does NOT have
available historical information or is NOT applicable Examples:
According to Ross: - Stock price, Sensex
Judgemental - Exchanging rate, interest rate, inflation rate,
- Expert opinion National GDP
- Sales force estimate - Retail Sales
- Historical Analogy - Electric Power Consumption
- Panel Consensus - Number of Accident Fatalities
- Market Research
 Time Series Patters Doane and Seward (2016) said that:
Anderson et. al.,(2018) stated the common types of - Seasonal (S) is a repetitive cyclical pattern
data patterns that can be identified when within a year (or within a week, day, or
examining a time series plot include: other time period).
- Over a small number of time periods, cycles
1. Horizontal (H) are undetectable or may resemble a trend.
Exists when the data values fluctuate - By definition, annual data have no
around a constant mean or stationary seasonality.

2. Trend (T)
-Trend (T) is the general movement over all 4. Cyclical (C)
years (t = 1, 2, ..., n). Exists when the data exhibit rises and falls
-A mathematical trend can be fitted to any that are not a fixed period.
data but may or may not be useful for
predictions.
-Trends may be steady and predictable,
increasing, decreasing, or staying the same.

- A cyclical pattern exists if the time series

plot shows an alternating sequence of
points below and above the trend line
lasting more than one year.
- Often, the cyclical component of a time
3. Seasons (S)
series is due to multiyear business cycles.
Exists when a series is influenced by
- Business cycles are extremely difficult, if not
seasonal factors or in regular interval.
impossible, to forecast.
(ex. The quarter of the year, the month or
- In this chapter we do not deal with cyclical
day of the week)
effects that may be present in the time
series.
Doane and Seward illustrated that:
- Cycle (C) is a repetitive up-and-down
movement about a trend that covers
several years.
Trend and Seasonal Pattern - Over a small number of time periods, cycles
- Some time series include a combination of a are undetectable or may resemble a trend.
trend and seasonal pattern.
- In such cases we need to use a forecasting
method that has the capability to deal with
both trend and seasonality.
- Time series decomposition can be used to
separate or decompose a time series into
trend and seasonal components.
Rahman (2015) reported the following: Forecast error
Uses of Time Series: - difference between the actual value and the
- To study the past behavior of variable value forecasted for a given period.
- To formulate policy decisions and planning
of future operations Actual Value – Forecast = Forecast Error
- To predict or estimate or forecast the - A positive forecast error indicates the
behavior of the phenomenon in future forecasting method underestimated the
which is very essential for business planning actual value.
- To compare the changes in the values of ✔ A negative forecast error indicates the
different phenomenon at different values forecasting method overestimated the
actual value.
Anderson et al., (2018) Averred that…
• The underlying pattern in the time series is Forecast Accuracy: General Idea
an important factor in selecting a
forecasting method.
• Hence, if we observe a trend in the data,
then we need to use a method that has the
capability to handle trend effectively.
• To select which forecasting method (which
will be discussed in the following chapters) Mean Absolute Deviation (MAD)
is appropriate or effective, we need to - A MAD of 0 indicates the forecast exactly
measure the accuracy of the forecast. Thus, predicted demand.
the following slides will focus on various Mean Absolute Percentage (MAPE)
ways of evaluating forecast accuracy with - Provides a perspective of the true
examples. magnitude of forecast error
Mean Squared Error (MSE)
Forecast Accuracy: Concepts, Terms and - Analogous to variance, large forecast errors
Definitions are heavily penalized.
-
Anderson (2018) defined the following..
How to measure forecast accuracy? (Aderson et.
• Measures of forecast accuracy are used to al, 2018)
determine how well a particular forecasting - To demonstrate the computation of these
method is able to reproduce the time series measures of forecast accuracy we will
data that are already available. introduce the simplest of forecasting
• Measures of forecast accuracy are methods.
important factors in comparing different - The naïve forecasting method uses the most
forecasting methods. recent observation in the time series as the
• By selecting the method that has the best forecast for the next time period.
accuracy for the data already known, we
hope to increase the likelihood that we will
obtain better forecasts for future time
 
periods. - forecast accuracy may not assure of good
forecasts.
- forecast accuracy gives an idea of how close
Forecast bias the forecasted values, predicted by the
persistent tendency for forecasts to be greater or estimated model, or be alike with the actual
less than the actual values of a time series.
 values of the corresponding historical past
data.
- MAD computes absolute difference.
- MAPE computes relative to the totality of
the data and commonly used when
comparing cases of different size or
degree/amount. (e.g case 1 in millions; case
2 in tens or units)
- ** the lower the value of any of the 3
measures, the better quality of forecast
accuracy.