Time Series Analysis
Time Series Analysis is a statistical technique used to model and analyze a
sequence of data points collected or recorded at regular time intervals. The
primary goal of time series analysis is to understand the underlying patterns
in the data over time and make forecasts for future points in the series.
Time series analysis is a powerful tool for examining data points that are
collected sequentially over time, such as stock prices, weather data, sales
figures, or economic indicators.
A time series can be univariate (one variable measured over time) or
multivariate (multiple variables).
A series of observations are recorded in accordance with the time of
occurrence is called Time Series.
Such data are particularly interest in Economics, Business and Commerce
Eg : Production ,Consumption, Sales, Profits,etc
Symbolically , Yt denotes the value of the variable at time ‘t’(t=1,2,3,4…,n)
Components of Time Series Analysis:
A graphical representation of time series will
reveal the changes over time.(horizontal line
means no change)
Components:
1. Trend (long term movement)
2. Seasonality Trend
3. Cyclic Trend
4. Random/Irregular/Noise movement
Trend (T)
-it is smooth, regular and long term movement exhibiting the tendency of
growth or decline over a period of time.
The trend represents the general direction in which the time series data is
moving over a long period. Trends can be upward, downward, or stationary,
depending on whether the data is increasing, decreasing, or remaining
stable over time.
Upward Trend: When data points increase consistently over time. This
could be caused by economic growth, technological improvements,
etc.
o Example: Long-term growth in stock prices or a company’s
revenue.
Downward Trend: When data points decrease consistently over time.
o Example: Decline in sales of physical books due to the rise of e-
books.
Stationary Trend: When there is no significant long-term upward or
downward movement.
o Example: A company's stable production output over time.
The trend can be linear (constant slope) or non-linear (the slope changes
over time). Non-linear trends often require more complex models to capture
changes.
2. Seasonality (S)
The component responsible for regular rise and fall in the magnitude of the
time series is called seasonal trend/variation
It refers to periodic variation in time series which occurs at regularly within
a period of 12 months.
Seasonality refers to short-term, periodic fluctuations that occur at regular
intervals due to seasonal factors. The key aspect of seasonality is that the
pattern repeats at a fixed frequency, such as daily, monthly, or yearly.
Examples of Seasonality:
o Retail sales tend to increase every December due to the holiday
shopping season.
o Electricity consumption peaks during summer and winter due to
the use of air conditioning and heating, respectively.
Seasonality is essential in time series forecasting, as ignoring seasonal
patterns can lead to inaccurate predictions.
3. Cyclic Variation (C)
Cyclicality refers to long-term oscillations in the data that occur due to
business or economic cycles. These cycle may or may not be periodic i.e.
they may or may not follow exactly similar pattern after equal interval of
time. Cycles are different from seasonal effects because cycles do not have a
fixed or predictable frequency, and their duration may vary.
Characteristics of Cyclic Patterns:
o Cycles can be affected by economic expansions and recessions.
o Unlike seasonal fluctuations, cycles can span over several years.
Example of Cyclicity:
o Economic indicators such as GDP growth follow a cyclical pattern
due to periods of boom and bust in the economy.
Cyclicality is often harder to model and forecast because of its unpredictable
nature. However, econometric models and certain smoothing techniques are
commonly used to capture these effects.
4. Irregular Component (Noise/Random Variation)
The irregular component, also known as noise or random variation,
represents unpredictable, random fluctuations in the time series data. These
variations are caused by unpredictable or unforeseen events that cannot be
attributed to the trend, seasonality, or cycles.
Examples of Irregular Component:
o Sudden drop in stock prices due to an unexpected event like a
natural disaster or political upheaval.
o A one-time spike in sales due to a celebrity endorsement.
The irregular component is the residual after all other components (trend,
seasonality, and cyclicality) have been accounted for. Random variations are
usually non-repeating and do not exhibit any systematic behavior.
Analysis of Time Series : The main objective of time series analysis is to give
the mathematical description of the component variations present. In order
to meet this objectives, it is necessary to break down the series into its main
components namely, trend, seasonality , cyclic and random movement.
Decomposition of Time Series
Time Series Models:
To better understand the time series, it is often decomposed into its
constituent components. There are two types of decomposition:
1. Additive Decomposition/Model: When the components are assumed
to be additive, the time series can be represented as:
Yt=T+S+C+I
where:
o Yt is the observed value of variable at time t,
o T is the trend component,
o S is the seasonal component,
o C is the cyclic component, and
o I is the irregular or noise component.
Additive decomposition is suitable when the seasonal variations or cycles do
not depend on the level of the series (i.e., they remain constant over time).
2. Multiplicative Decomposition: When the components are assumed to
be multiplicative, the time series can be represented as:
Yt=T*S*C*I
Multiplicative decomposition is useful when the magnitude of seasonal or
cyclic variations grows or shrinks depending on the trend level.
Utility of Time Series Analysis:
The time series analysis is useful in many areas such as Business,Economics,
Commerce ,demography for the reason given below:
1. It helps in evaluation of current accomplishments
2. It gives general description of past behavior of series
3. It helps in forecasting the future behavior on the basis of past behavior
4. It facilitates comparison
Measurement of Trend:
Graphical Method
Moving Average method
Least square method
Semi average method
Graphical Method
The simplest method to measure a trend is by plotting the data on a graph
(usually a line chart) and visually inspecting it for any patterns.
How to Apply: Plot the time series data with time on the x-axis and the
variable of interest on the y-axis. Look for an upward, downward, or
stationary movement over time.
Limitation: This method is subjective and not very precise, especially
when the data has high variability or noise.
