Unit-3

Part-A

Apr-2024

1.State the purpose of adding additional quantitative and /or categorial explanatory variables to any
developed linear regression model.Justify with an example.

Adding additional quantitative or categorical explanatory variables to a linear regression model serves
several important purposes:

1. Improved Model Fit: Including more relevant variables can capture additional variation in the
response variable, leading to a better fit. This reduces residual variance and improves the
model's explanatory power.

2. Control for Confounding: Additional variables can help control for confounding factors that
might bias the estimated effects of primary predictors. This leads to more accurate estimates
and inferences.

3. Enhanced Predictive Accuracy: More variables can improve the model's predictive accuracy
when applied to new data, particularly if the additional variables are related to the response.

4. Identification of Interactions: Including categorical variables allows for the exploration of

interaction effects between predictors, providing insights into how different factors may
combine to affect the outcome.

Example
Suppose we are developing a linear regression model to predict house prices based on square footage.
Initially, we might have a model like this:

While square footage is a significant predictor, it might not capture all the factors influencing house
prices. We could enhance the model by adding categorical variables such as neighborhood and the
presence of a garage.

Justification
1. Improved Model Fit: By adding "Neighborhood," we account for the fact that houses in different
areas may sell for different prices regardless of size. This variable can significantly improve the
model fit by capturing location-specific price trends.

2. Control for Confounding: Including "Garage" helps control for the additional value associated
with houses that have garages, which might be correlated with square footage but is also an
independent factor influencing price.

3. Enhanced Predictive Accuracy: The enhanced model can better predict prices in new data by
considering these additional relevant factors, resulting in a more reliable estimation of house
values.
 4. Identification of Interactions: If we suspect that the effect of square footage on price varies by
neighborhood, we could include an interaction term (e.g., Square Footage × Neighborhood),
allowing for a more nuanced understanding of how these variables interact.

In summary, adding more explanatory variables helps create a more comprehensive model that
captures the complexity of real-world relationships, leading to more accurate and reliable predictions.

2.Give an example of a data set with a non-Gaussian distribution.

Certainly! One common example of a non-Gaussian distribution is the exponential distribution. This
distribution is often used to model the time between events in a Poisson process, such as the time until
a radioactive particle decays or the time until a customer arrives at a store.

Example Dataset
Here’s a small example dataset that follows an exponential distribution with a mean (λ) of 1:

Observation
0.4

1.1

0.1

2.0

0.7

1.5

0.9

0.3

1.2

0.5

Characteristics
 Shape: The exponential distribution is right-skewed, meaning it has a long tail on the right.
 Mean and Variance: The mean is equal to the standard deviation in the exponential distribution.

This dataset is a great example of how non-Gaussian distributions can manifest in real-world scenarios!
Nov-2023

3.Define multiple regressions.

Multiple regression is a statistical technique used to understand the relationship between one
dependent variable and two or more independent variables. It helps to assess how well the independent
variables explain variations in the dependent variable.

In a multiple regression model, the dependent variable is predicted based on the linear combination of
the independent variables, each multiplied by a coefficient that represents the strength and direction of
the relationship. The general form of the equation is:

Y=β0+β1X1+β2X2+...+βnXn+ϵ

Where:

 YYY is the dependent variable.

 β0\beta_0β0 is the intercept.
 β1,β2,...,βn are the coefficients for the independent variables X1,X2,...,Xn
 Ε epsilonϵ is the error term, representing the variation in YYY not explained by the independent
variables.

Multiple regression is useful for making predictions, testing hypotheses about relationships, and
controlling for the effects of other variables.

4.Define regression towards the mean.

Regression toward the mean is a statistical phenomenon that occurs when an extreme observation is
followed by a subsequent observation that is closer to the average or mean. This concept suggests that
extreme values are often followed by values that are less extreme. For example, if a student scores
exceptionally high on a test one year, their score may be lower the next year, moving closer to the
average score of their peers. This effect arises because random variation can lead to extreme outcomes,
but the inherent variability in measurements or performances often results in more typical results over
time.

Apr-2023

5.what is the use of scatter plot?

A scatter plot is a graphical representation used to display the relationship between two quantitative
variables. Here are some key uses:

1. Visualizing Relationships: It helps identify correlations or patterns between variables, such as

positive, negative, or no correlation.
 2. Identifying Trends: You can observe trends in data, like linear or non-linear relationships, which
can inform further analysis.

3. Outlier Detection: Scatter plots make it easy to spot outliers or anomalies in the data that might
need further investigation.

4. Comparing Groups: You can use different colors or markers to represent different categories or
groups, allowing for easy comparison.

5. Regression Analysis: They serve as a preliminary tool for assessing the suitability of regression
models.

6.Define correlation coefficient.

The correlation coefficient is a statistical measure that quantifies the degree to which two variables are
related. It ranges from -1 to 1, where:

 1 indicates a perfect positive correlation, meaning that as one variable increases, the other also
increases.
 -1 indicates a perfect negative correlation, meaning that as one variable increases, the other
decreases.
 0 indicates no correlation, meaning there is no linear relationship between the variables.

The most commonly used correlation coefficient is Pearson's correlation coefficient, which assesses the
strength and direction of a linear relationship. Other types, like Spearman's rank correlation, measure
relationships that may not be linear.

Nov-2022

7.consider Helen sent 10 greeting cards to her friends and she received back 8 cards, what is the kind of
relationship it is? Brief on it.

The relationship between Helen and her friends can be described as reciprocal or mutual. In this
context, Helen sent out 10 greeting cards, which indicates her effort to connect and maintain
relationships. The fact that she received back 8 cards suggests that her friends appreciated her gesture
and reciprocated it, showing they value the relationship as well. This exchange highlights a give-and-take
dynamic, where both parties contribute to and nurture their connections.

Part-B

Apr-2024

1.a)i)In statistics , highlight the impact when the goodness of fit test score is low.
A low goodness of fit test score indicates that the observed data significantly deviates from the expected
model. Here are some key impacts:

1. Model Mis-specification: A low score suggests that the chosen model may not accurately
capture the underlying data patterns, indicating the need for a different model.

2. Increased Error Rates: Predictions made from a poorly fitting model are likely to have higher
errors, leading to unreliable conclusions or forecasts.

3. Data Interpretation Challenges: A low score complicates the interpretation of results, as it raises
questions about the validity of any inferences drawn from the model.

4. Assumption Violations: It may signal that assumptions underlying the statistical methods (e.g.,
independence, normality) are violated, which could further skew results.

5. Model Complexity: A low score could lead to overfitting if more complex models are used to
achieve a better fit, potentially harming generalizability.

6. Need for Data Transformation or Additional Variables: It may highlight the necessity for data
transformations, interactions, or additional variables to improve fit.

7. Re-evaluation of Research Questions: It might prompt a re-assessment of the research

questions and hypotheses to ensure they align with the observed data.

In summary, a low goodness of fit score is a critical signal that should lead to further investigation and
potential adjustments in modeling strategy.

ii)Given the following dataset of employee, Using regression analysis, find the expected salary of an
employee if the age is 45.

Age 54 42 49 57 35
Salary 67000 43000 55000 71000 25000

Step 1: Calculate the Regression Line

We can use the formula for a linear regression model:

y=mx+by = mx + by=mx+b

Where:

 yyy is the salary,

 xxx is the age,
 mmm is the slope of the line,
 bbb is the y-intercept.
b)i)Define auto correlation and how is it calculated?what does the negative correlation convey?

Autocorrelation is a statistical measure that describes the correlation of a time series with its own past
values. It helps to identify patterns, trends, and seasonality in the data, indicating how current values
are related to their previous values.

Calculation of Autocorrelation
1. Mean Calculation: First, compute the mean of the time series.
2. Lagged Values: Select a lag (k) and create lagged versions of the time series (shift the series by k
periods).
3. Covariance Calculation: Calculate the covariance between the original series and the lagged
series.
4. Variance Calculation: Compute the variance of the original time series.
5. Autocorrelation Formula:
Negative Autocorrelation
A negative autocorrelation indicates that if a value is high (or low), the subsequent value is likely to be
low (or high). This can signify a potential oscillating pattern or a mean-reverting process in the data.
Negative autocorrelation is often seen in phenomena like financial markets, where high returns may be
followed by low returns, suggesting a tendency to revert to the mean.

ii)what is the philosophy of logistic regression what kind of model it is?what does logistic regression
predict?Tabulate the cardinal differences of Linear and Logistic regression.

Logistic regression is a statistical model used primarily for binary classification tasks, where the outcome
variable is categorical (typically taking on two values, such as 0 and 1). Here’s a brief overview of its
philosophy, model characteristics, predictions, and differences from linear regression:

Philosophy of Logistic Regression

Logistic regression models the probability that a given input belongs to a particular category. Unlike
linear regression, which predicts a continuous output, logistic regression uses the logistic function
(sigmoid function) to constrain the predicted probabilities to the range [0, 1]. This makes it well-suited
for binary outcomes.

Type of Model
 Type: Logistic regression is a type of generalized linear model (GLM).
 Link Function: It uses the logit link function, which relates the linear combination of input
features to the odds of the outcome occurring.

What Does Logistic Regression Predict?

Logistic regression predicts the probability that an instance belongs to the positive class (typically
represented as 1). The output is transformed using the logistic function to ensure it lies between 0 and
1. A threshold (commonly 0.5) is applied to make a classification decision.
Cardinal Differences Between Linear and Logistic Regression

Summary
Logistic regression is a powerful tool for binary classification problems, leveraging the relationship
between input features and the log-odds of the outcome. Understanding its distinct characteristics
compared to linear regression is crucial for choosing the right model for your data.

Nov-2023

2)a)i)Explain scatter plot

A scatter plot is a type of data visualization that uses dots to represent the values obtained for two
different variables—one plotted along the x-axis and the other along the y-axis. Each dot on the plot
corresponds to an observation in the dataset, with its position determined by the values of the two
variables.

Key Features of a Scatter Plot:

1. Axes:

 The horizontal axis (x-axis) represents one variable.

 The vertical axis (y-axis) represents the second variable.
2. Data Points:
  Each point represents an individual observation, with its coordinates determined by the
values of the two variables.
3. Trends and Patterns:

 By examining the arrangement of the points, you can identify relationships, such as
positive correlation (both variables increase together), negative correlation (one variable
increases while the other decreases), or no correlation (no discernible pattern).
4. Outliers:

 Points that are significantly distant from the rest of the data can indicate outliers, which
may warrant further investigation.
5. Clustering:

 Groupings of points can indicate clusters or categories within the data.

Applications:
 Exploratory Data Analysis: Helps identify potential relationships between variables before more
complex analyses.
 Regression Analysis: Assists in visualizing relationships that can be modeled with statistical
techniques.

Example:
If you were to plot the height of individuals on the x-axis and their weight on the y-axis, each point
would represent a different person. You might observe a trend where taller individuals tend to weigh
more, indicating a positive correlation between height and weight.

Overall, scatter plots are a powerful tool for visualizing and understanding the relationships between
two quantitative variables.

ii)Describe range and variance

Range
The range is a measure of the spread of a dataset. It is calculated by subtracting the smallest value from
the largest value. The formula is:

Range=Maximum Value−Minimum Value\text{Range} = \text{Maximum Value} - \text{Minimum

Value}Range=Maximum Value−Minimum Value

For example, in the dataset [3,7,2,5][3, 7, 2, 5][3,7,2,5]:

 Maximum = 7
 Minimum = 2
 Range = 7−2=5
Variance
Variance measures how much the values in a dataset differ from the mean (average) of the dataset. It
quantifies the degree of spread or dispersion. The formula for variance (for a sample) is:

b)i)Explain the correlation

coefficient.

ii)Explain how the least

squares equation which is
used to minimize the total
of all squared prediction
errors with example.

Apr-2023

3.a)calculate the correlation coefficient for the heights ‘in inches’ of father’s(x) and their son’s(y) with
the data presented below.

X 66 68 68 70 71 72 72
Y 68 70 69 72 72 72 74

To calculate the correlation coefficient (Pearson's rrr) for the heights of fathers (X) and their sons (Y), we
can use the following formula:

Given Data:
 XXX (Father's heights): 66, 68, 68, 70, 71, 72, 72
 YYY (Son's heights): 68, 70, 69, 72, 72, 72, 74

Step 1: Calculate sums and products

1. Count n: The number of pairs is 7.
Step 2: Plug values into the formula
Now, substituting the values we found into the correlation formula:

Conclusion
After performing the calculation:

The correlation coefficient r is approximately 0.86, indicating a strong positive correlation between the
heights of fathers and sons.
b)The value of x and their corresponding values of y are presented below.

X .5 1.5 2.5 3.5 4.5 5.5 6.5

Y 2.5 3.5 5.5 4.5 6.5 8.5 10.5

i)find the least square regression line y=ax+b

ii)Estimate the value of y when x=10

To find the least squares regression line y=ax+by

and to estimate the value of y when x=10, we can follow these steps:

Step 1: Calculate the necessary sums

Given the values:

X:0.5,1.5,2.5,3.5,4.5,5.5,6.5

Y:2.5,3.5,5.5,4.5,6.5,8.5,10.5

Let’s calculate the following sums:

 n (number of data points)

Step 2: Perform the calculations

1. Number of data points (n):

n=7
Step 3: Calculate coefficients a and b
Now we can use the formulas for the slope a and intercept b:
Nov-2022

4.a)i)categorize the different types of relationships using scatter plots.

Using scatter plots to categorize different types of relationships is a visual way to represent correlations
between two variables. Here’s how you can categorize various relationship types based on the patterns
observed in scatter plots:

1. Positive Linear Relationship

 Description: As one variable increases, the other variable also increases.
 Scatter Plot Appearance: Points trend upwards from left to right.
 Example: Height vs. weight.

2. Negative Linear Relationship

 Description: As one variable increases, the other variable decreases.
 Scatter Plot Appearance: Points trend downwards from left to right.
 Example: Age vs. speed of reaction.

3. No Relationship (Random Scatter)

 Description: No discernible pattern or correlation between the variables.
 Scatter Plot Appearance: Points are randomly distributed.
 Example: Shoe size vs. intelligence.

4. Curvilinear Relationship
 Description: The relationship between variables follows a curved line, indicating that the rate of
change is not constant.
 Scatter Plot Appearance: Points form a U-shape or an inverted U-shape.
 Example: Stress level vs. performance (Yerkes-Dodson Law).

5. Strong Correlation
 Description: A clear pattern exists where the points are closely clustered around a line or curve.
 Scatter Plot Appearance: Points are tightly packed around a linear or curvilinear trend.
 Example: Temperature vs. ice cream sales.

6. Weak Correlation
 Description: A pattern exists, but the points are more spread out.
 Scatter Plot Appearance: Points are loosely arranged around a trend line.
 Example: Study hours vs. exam scores (with some variability).

7. Outliers
 Description: Data points that lie far away from the overall pattern, which can significantly
influence the correlation.
  Scatter Plot Appearance: Isolated points that do not follow the general trend.
 Example: A student who studies very little but scores exceptionally high on a test.

Visualization Tips
 Axes Labels: Clearly label the axes with the variables being compared.
 Trend Lines: Use regression lines to illustrate the relationship type.
 Color Coding: Different colors can represent different types of relationships for comparative
purposes.

Using these categories, you can effectively interpret and communicate the nature of relationships
observed in data.

ii)Each of the following pairs represents the number of licensed drivers(X) and the number of cars(Y) for
seven houses in my neighborhood:

Drivers 5 5 2 2 3 1 2
(X)
Cars(Y) 4 3 2 2 2 1 2
1)construct a scatterplot to verify a lack of pronounced curvilinearity.

2)Determine the least squares equation for these data.(Remember, you will first have to calculate r, SSy
and SSx)

Refer class work out

3)Determine the standard error of estimate, Sy/x,given that n=7.

b)i)In studies dating back over 100 years, it’s well established that regression toward the mean occurs
between the heights of fathers and the heights of their adult Sons.

Indicate whether the following statements are true or false.

1)sons of tall fathers will end to be shorter than their fathers.

2)sons of short fathers will tend to be taller than the mean for all sons.

3)Every son of a tall father will be shorter than his father.

4)Taken as a group, adult sons are shorter than their fathers.

5)fathers of tall sons will tend to be taller than their sons.

6)fathers of short sons will tend to be taller than their sons but shorter than the mean for all fathers.

ii)Interpret the value of r 2 in correlation based analysis.

i) True or False Statements
1. False: Sons of tall fathers will tend to be shorter than their fathers, but this doesn't mean they
will always be shorter; they just regress toward the mean height.

2. True: Sons of short fathers will tend to be taller than the mean for all sons, as they regress
toward the mean.

3. False: Not every son of a tall father will be shorter than his father; some may still be taller, but
the average will be shorter.

4. False: Taken as a group, adult sons are not shorter than their fathers; they tend to be closer to
the average height.

5. True: Fathers of tall sons will tend to be taller than their sons, as the sons regress toward the
mean.

6. True: Fathers of short sons will tend to be taller than their sons but shorter than the mean for all
fathers, again due to regression toward the mean.

ii) Interpretation of the Correlation Coefficient (r)

The correlation coefficient (r) ranges from -1 to 1 and indicates the strength and direction of a linear
relationship between two variables.

 r = 1: Perfect positive correlation; as one variable increases, the other variable also increases.
 r = -1: Perfect negative correlation; as one variable increases, the other decreases.
 r = 0: No correlation; changes in one variable do not predict changes in the other.
 Values close to 1 or -1 indicate a strong relationship, while values close to 0 indicate a weak
relationship.

In the context of regression toward the mean, a correlation close to 0 would suggest that extreme
heights (either tall or short) in fathers are not strong predictors of their sons' heights, as the sons will
tend to regress toward the average.