SCRIPT: SLIDE 2

SLIDE 1

Good afternoon everyone!

Today, we’re going to explore an essential

statistical technique that plays a crucial role
in data analysis, business forecasting,
machine learning, and many other fields—
Regression Analysis.

So, what exactly is regression analysis? Now, let’s talk about the importance of
correctly identifying the dependent and
independent variables in correlation and
regression analysis.

When analyzing data, it’s crucial to

distinguish between these two variables
because misidentifying them can lead to
incorrect conclusions.

The independent variable (X) is the factor

that influences or predicts another variable,
while the dependent variable (Y)is the
At its core, regression is a powerful tool that outcome that depends on changes in X. A
helps us understand relationships between simple way to remember this is: X causes
variables. It allows us to examine how one the effect, and Y is the effect.
variable, known as the dependent variable,
changes in response to one or more If we mix these up, our scatter plots might
independent variables. In simple terms, it not make sense, and the correlation
helps us make predictions and identify coefficient could become misleading. For
patterns in data. example, if we study the relationship
between study hours and exam scores, the
For example, businesses use regression to correct setup is that study hours (X)
predict sales based on marketing expenses, influence exam scores (Y). If we reverse
economists analyze how interest rates them, it wouldn't make logical sense because
impact inflation, and scientists use it to exam scores don’t determine how much a
model population growth or climate change student studies.
trends. Whether we’re dealing with simple
linear relationships or more complex Another example is temperature and ice
patterns, regression gives us valuable cream sales. Warmer temperatures lead to
insights into the connections between data higher ice cream sales, not the other way
points. around. If we swap these variables, we
might draw the wrong conclusions.

1
So, to ensure accurate and meaningful making them a key tool in regression and
analysis, always double-check that you’ve correlation analysis.
correctly identified your independent and
dependent variables before interpreting SLIDE 12
results.

SLIDE 11

Now, why is a scatter plot so important in

regression analysis?

A scatter plot is a type of graph that helps Scatter plots play a crucial role in regression
us visualize the relationship between two analysis because they allow us to visually
variables. It displays individual data points assess the relationship between two
on a two-dimensional plane, where: variables before performing any
calculations. Here’s why they matter:
• The x-axis represents
the independent variable (the 1. Visualizing Relationships – A
predictor or cause). scatter plot helps us determine if
• The y-axis represents the dependent there is a relationship between the
variable (the outcome or effect). variables. It can show whether the
• Each dot on the graph represents one relationship is linear, nonlinear, or
observation or data point. if there’s no correlation at all. If
the points form a clear pattern, we
By looking at a scatter plot, we can know a regression model may be
quickly identify patterns or trends in the useful.
data. For example, if the points form an 2. Identifying Patterns – Scatter plots
upward trend, we may have a positive reveal important patterns like trends,
correlation—meaning as X increases, Y clusters, or gaps in data.
also increases. If they form a downward Recognizing these patterns can help
trend, we see a negative correlation— improve the accuracy of a regression
where X increases and Y decreases. And model.
if the points are scattered randomly, 3. Assessing Linearity – Regression
there may be no correlation between the models often assume
variables. Scatter plots are powerful a linear relationship between
because they allow us to visually assess variables. By looking at a scatter
relationships before performing any plot, we can check if the data follows
calculations. They help us spot trends, a straight-line trend or if a different
outliers, and possible errors in data, approach, like polynomial
regression, might be needed.

2
 4. Detecting Outliers – Outliers are relationship between study hours and
data points that don’t follow the test scores—more study time
general trend and generally leads to higher scores.
can skew regression results. Scatter 2. Negative Relationship – If the
plots make it easy to spot and points trend downward from left to
investigate these unusual points right, it shows a negative
before running the analysis. correlation. This means that as X
5. Checking the Strength of increases, Y decreases. An example
Association – The way the points are of this would be the relationship
clustered gives us an idea of between the number of absences in
the strength and direction of the class and exam scores—more
relationship. If the points are closely absences usually result in lower
packed along a trend, we know scores.
there’s a strong correlation. If they’re 3. Curvilinear Relationship –
spread out, the relationship is Sometimes, the data doesn’t follow a
weaker. straight-line pattern. Instead, it
curves upward or downward,
Overall, scatter plots provide a quick, indicating a nonlinear relationship.
intuitive way to evaluate data before diving For example, the relationship
into regression calculations, making them between stress and performance—at
a critical first step in regression analysis. low levels, stress can improve
performance, but too much stress can
SLIDE 13 reduce it, forming a curved pattern.
4. No Relationship – If the points
are scattered randomly with no
clear pattern, it suggests that there
is no correlation between the two
variables. In this case, changes in X
do not predict changes in Y. An
example might be someone’s shoe
size and their intelligence—there’s
no meaningful connection between
the two.

As you can see, scatter plots can reveal By looking at these patterns in a scatter
different types of relationships between plot, we can determine the nature of the
two variables. Let’s go through the main relationship between variables, which
types one by one." helps us decide on the right regression
model to use.
1. Positive Relationship – When the
points on the scatter plot
trend upward from left to right, it
indicates a positive correlation. This
means that as the independent
variable (X) increases, the
dependent variable (Y) also
increases. A common example is the

3
SLIDE 14 2. The magnitude of the change –
This tells us how much Y changes
for every one-unit increase in X.

A steeper slope indicates a larger change in

Y for each change in X.

A flatter slope means that Y changes less in

response to X.

However, it’s important to note that while

the slope tells us the rate of change, it
The predicted value of Y is equal to the does not indicate how strong the
intercept (β) plus the slope (β1) multiplied relationship is. The strength of the
by the independent variable (X) relationship is determined by other
measures, such as the correlation
SLIDE15 coefficient (r) or R-squared value, which
assess how well the independent variable
explains the variation in the dependent
variable.

So, while the regression coefficient helps

us understand the effect of X on Y, we
need to look at additional statistical
measures to fully interpret the strength of
the relationship.

Now, let’s talk about the regression

coefficient, which is one of the key POLYNOMIAL
components of regression analysis."
SLIDE 16
The regression coefficient, also known as
the slope, tells us two important things:

1. The direction of the relationship –

Whether the dependent variable (Y)
increases or decreases as the
independent variable (X) changes.

A positive slope means that as X increases,

Y also increases.

A negative slope means that as X increases, Now, let’s talk about polynomial
Y decreases. regression and how it’s different from
simple linear regression.

4
In simple linear regression, we use
a straight line to show the relationship
between two variables. But sometimes, data
doesn’t follow a straight-line pattern—it
curves. That’s where polynomial
regression comes in.

Polynomial regression allows us to fit a

curved line to the data instead of a straight
one. It does this by adding powers of
X (like X2, X3, etc.) to the equation.

Equation: Polynomial regression builds on linear

regression by adding higher powers of the
independent variable, which allows us to
fit curves instead of just straight lines."

The degree (n) of the

polynomial determines how flexible the
This is useful when data follows a U-shape, model is:
an S-shape, or any curved pattern, like
predicting population growth, temperature • A lower-degree
changes, or the path of a ball in motion. polynomial (like X2X2) captures
simple curves.
In short, polynomial regression helps us • A higher-degree
model real-world relationships that aren’t polynomial (like X5X5 or X6X6)
just straight lines. can fit more complex patterns in the
data.
SLIDE 17
However, adding too many polynomial
terms can lead to overfitting, where the
model becomes too sensitive to small
fluctuations in the data rather than capturing
the overall trend.

So, while polynomial regression helps us

model curves, it's important to find the
right balance to avoid overfitting.

SLIDE 19
SLIDE 18

5
 5. No Multicollinearity – This is
especially important in polynomial
regression because when we
introduce higher-degree terms
(like X2,X3,X4X2,X3,X4), they can
become highly correlated with each
other. High multicollinearity can
make it difficult to determine the true
effect of each term.

Just like linear regression, polynomial By checking these assumptions, we can

regression relies on a few key assumptions ensure that our polynomial regression
to ensure accurate results. Let’s go model is reliable and provides meaningful
through them one by one." insights.

1. Linearity (in terms of

coefficients) – Even though
polynomial regression models
curves, it is still considered a linear
model in terms of the
coefficients (β values). This means
the equation remains a sum of terms
rather than involving multiplication
of coefficients.
2. Independence of Errors – The
errors (or residuals) should
be independent, meaning that one
prediction error should not be
influenced by another. If errors are
correlated, it can lead to misleading
results.
3. Homoscedasticity – This means that
the spread of residuals should be
roughly the same across all values of
the independent variable. If the
spread changes (for example, if
errors get larger as X increases), it
can indicate a problem in the model.
4. Normality of Errors – The errors
should follow a normal
distribution. This assumption helps
ensure that confidence intervals and
hypothesis tests are valid. A quick
way to check this is by looking at a
histogram or a normal probability
plot of the residuals.