Regression Analysis
Part 01 – Technology and Analy4cs
SECTION F
Analy&c Tools
Linear regression analysis is a sta&s&cal method used to model the rela&onship
between a dependent variable and one or more independent variables. The goal is
to understand and predict the behaviour of the dependent variable based on the
values of the independent variable.
Time series analysis involves analysing data points collected or recorded at
specific &me intervals. This method aims to understand pa;erns, trends, and
behaviours within the data over &me. Time series data typically includes
observa&ons taken at equally spaced intervals, such as hourly, daily, monthly, or
yearly.
Four components of &me series
1. Trend pa;ern or component
2. Cyclical pa;ern or component
3. Seasonal pa;ern or component
4. Irregular pa;ern or component
Regression Analysis
1) Trend Pa3ern: The long-term movement or
direc&onality in the data. Trends can be upward
(increasing), downward (decreasing), or stable.
Trend Pa1ern
Trend Pattern
2) Cyclical pa3erns in &me series
analysis refer to recurring
fluctua&ons or movements that
occur at irregular intervals over an
extended period. These cycles oLen
represent economic, business, or
societal trends that repeat over a
more extended period than
seasonal pa;erns. Economic cycles
like expansions and contrac&ons,
business cycles, and various societal
trends are examples of cyclical
pa;erns.
Cyclical Pa1ern
Cyclical Pa1ern
3) Seasonal pa3erns in &me series
analysis refer to regular fluctua&ons
or varia&ons that occur at fixed
intervals within a specific period,
oLen within a year. These pa;erns
are typically influenced by factors
such as weather, holidays, customs,
or other calendar-related events.
Any pa;ern that repeats regularly is
a seasonal component.
Seasonal Pa1ern
Seasonal Pa1ern
4) Irregular Pattern
The irregular component in a time series
refers to the random fluctuations or
residual variations that cannot be
attributed to the underlying trend,
seasonal patterns, or cyclicality. Also
known as the "error" or "noise," this
component represents the unexplained
variability in the data after accounting for
the systematic patterns.
Irregular Pa1ern
Irregular Pa1ern
A &me series that has a long-term upward or downward trend pa;ern can be
used to make a forecast.
Simple linear regression analysis is used to create a trend projec&on and to
forecast values using historical informa&on from all available past observa&ons
of the value.
1. Simple Linear Regression
𝒚 = 𝒂 + 𝒃𝒙
2. MulAple Linear Regression
ŷ = 𝐚𝟎 + 𝐚𝟏 𝐱 𝟏 + 𝐚𝟐 𝐱 𝟐 + … + 𝐚𝐤 𝐱 𝐤
Regression Analysis
Correla'on Analysis
Used to understand the rela&onship or absence of a rela&onship between
two variables and to determine the strength of the linear rela&onship
between the two variables.
1) The correla(on coefficient, R
2) The coefficient of determina(on, 𝑹𝟐
3) The standard error of the es(mate, also called the standard error of the
regression
Regression Analysis
Correla'on coefficient (oLen denoted as R) is a sta&s&cal measure that
describes the strength and direc&on of a linear rela&onship between two
variables. It ranges from -1 to +1, indica&ng the strength and direc&on of the
rela&onship:
• R=1 represents a perfect posi&ve linear rela&onship: As one variable increases,
the other variable increases propor&onally.
• R=−1 represents a perfect nega&ve linear rela&onship: As one variable
increases, the other variable decreases propor&onally.
• R=0 indicates no linear rela&onship between the variables
ü high correla(on coefficient - when number closest to 1 or -1
ü moderate correla(on coefficient - ±0.30 to ±0.49
ü low correla(on coefficient - around ±0.10
Regression Analysis
The Coefficient of Determina'on
Percentage of the total varia(on in the dependent variable (y) that can be explained
by varia(ons in the independent variable (x), as depicted by the regression line.
• If R2 is 1, then 100% of the varia&on in the dependent variable is explained by
varia&ons in the
independent variable.
• If R2 is 0, then none of the varia&on in the dependent variable is explained by
varia&ons in the independent variable.
• If R2 is greater than 0 but less than 1, for example 0.68, it means that 68% of the
total varia&on in the dependent variable can be explained by varia&ons in the
independent variable.
Regression Analysis
The Standard Error of the Es'mate (SEE) or the Standard Error of
the Regression:
The equa&on of the simple linear regression model results in the average or predicted
value. The actual observed data has responses that are not on the line itself, but rather
they are sca<ered around the regression line.
The sca;er, that is, the difference between the actual value of the dependent variable y
and the predicted value of the dependent variable y (that is, ŷ) for each value of the
independent variable x, is called the error term or the residual for that value of x
y = dependent variable
a = constant coefficient
y = a + bx + e b = variable coefficient
x = independent variable
-
𝒆=𝒚−𝒚
e = error term
Regression Analysis
Trend Pa1ern
Goodness of Fit in Linear Regression Analysis
It describes how close the actual values used in a sta&s&cal model are to the
expected values, that is, the predicted values, in the model.
Low Goodness of Fit
Regression Analysis
High Goodness of Fit
Regression Analysis
Confidence Interval
The confidence interval is used to describe the amount of uncertainty caused by
the sampling method used when drawing conclusions about a popula&on based
on a sample. If several samples are drawn from a popula&on using the same
sampling method and a confidence interval at a confidence level of 95% is used,
95% of the interval es(mates in the samples can be expected to include the true
parameter of the popula(on.
Confidence Interval
Regression Analysis
• Rela(onship Iden(fica(on: It helps • To use regression analysis, historical
iden&fy and quan&fy rela&onships data are required. If historical data
between variables, allowing for the are not available, regression analysis
predic&on of one variable based on cannot be used.
others. • Even when historical data are
• Predic(on: Regression models can be available, the use of historical data is
used for forecas&ng and predic&ng ques&onable for making predic&ons if
future outcomes based on historical a significant change has taken place in
data. the condi&ons surrounding that data
Regression Analysis
RABEEH OVUNGAL
Thanks for Watching
Regression Analysis
Part 01 – Technology and Analytics
