A Project Report

on
SPSS Analysis using
Logistic Regression Data Set-2

Section: B

Group No.7

Submission to: Dr. Sunny Bose

NAME ENROLLMENT NO.

Harsh Sarin 23BSPHH01C0370

Dravyansh Vashishth 23BSPHH01C1422

Chaitanya Agrawal 23BSPHH01C0700

Ahmed Younus Bashoeb 23BSPHH01C1407

Vinjamuri Abishai Roy 23BSPHH01C1129

 SPSS ANALYSIS

Omnibus Tests of Model Coefficients

Chi-Square df Sig.

Step 1 Step 22.247 4 0.000

Block 22.247 4 0.000

Model 22.247 4 0.000

Model Summary

Step -2 Log likelihood Cox & Snell Nagelkerke

R Square R Square
1 13.643 0.575 0.768

Hosmer and Lemeshow Test

Step Chi-Square df Sig.

1 3.583 7 0.826

Contingency Table for Hosmer and Lemeshow Test

Success = N Success = Y
Observed Expected Observed Expected Total
Step 1 1 3 2.967 0 0.033 3
2 3 2.856 0 0.144 3
3 3 2.712 0 0.288 3
4 1 2.009 2 0.991 3
5 1 1.059 2 1.941 3
6 1 .336 2 2.664 3
7 0 .051 3 2.949 3
8 0 .008 3 2.992 3
9 0 .001 2 1.999 2
 Classification Table
Predicted
Observed Success Percentage
N Y Correct
Step 1 Success N 10 2 83.3
Y 2 12 85.7
Overall Percentage 84.6

Variables in the Equation

B S.E. Wald df Sig. Exp(B)

Step 1 CGPA 6.017 2.603 5.341 1 0.021 410.208
Specialization 0.928 2 0.629
Specialization (1) 1.598 1.898 0.709 1 0.400 4.943
Specialization (2) -0.317 1.946 0.027 1 0.871 0.728
Aptitude Score 0.510 0.305 2.802 1 .094 1.665
Constant -62.249 26.287 5.608 1 .018 0.000

Correlation Matrix

Specialization
Constant CGPA Specialization (1) (2) Aptitude Score
Step 1 Constant 1.000 -0.912 -0.373 0.028 -0.799
CGPA -0.912 1.000 0.456 -0.174 0.486
Specialization -0.373 0.456 1.000 0.330 0.058
(1)
Specialization .028 -0.174 .330 1.000 0.126
(2)
Aptitude Score -0.799 0.486 .058 .126 1.000
1. What is a logistic regression? How is it different from linear regression using
dummy variables?

Logistic regression is a statistical model used to predict the probability of a binary

outcome (e.g., yes/no, true/false, 0/1) based on one or more predictor variables.
It's a popular technique in machine learning and data analysis

Linear Regression with

Feature Dummy Variables Logistic Regression
Dependent
Variable Continuous Binary categorical
Prediction Continuous values Probabilities (0-1)
Transformation Dummy variables Logistic function
Represent the impact of
predictors on the dependent Represent the log-odds of the
Coefficients variable dependent variable being 1

2. Explain R square in the case of logistic regression?

R-squared is not directly applicable to logistic regression. While R-squared is a

commonly used metric in linear regression to measure the proportion of variance
explained by the model, it doesn't have a straightforward interpretation in logistic
regression. This is because the dependent variable in logistic regression is
categorical, not continuous.
Following are the reasons that R square does not work in case of Logistic
Regression-
• Non-linear relationship: Logistic regression models a non-linear relationship
between the predictors and the log-odds of the outcome. R-squared, which is
based on the linear relationship between predicted and actual values, doesn't
accurately capture this.
• Categorical dependent variable: R-squared assumes a continuous dependent
variable. In logistic regression, the dependent variable is categorical, making
R-squared inappropriate.
3. Interpret the Model Summary table.

• -2 Log Likelihood: This measures the overall fit of the model. Lower values
indicate a better fit. In this case, a value of 13.643 suggests a moderately good
fit.
• Cox & Snell R-squared: This is a pseudo R-squared measure that ranges
from 0 to 1. A higher value indicates a better fit. Here, 0.575 suggests a
moderate level of variance explained by the model.
• Nagelkerke R-squared: Another pseudo R-squared measure that can range
from 0 to 1. It often provides a more comprehensive assessment than Cox &
Snell R-squared. A value of 0.768 indicates a relatively strong fit.

4. Interpret the Variables in Equation table

Key columns:
• B: Unstandardized regression coefficient, representing the change in log-odds
for a one-unit increase in the predictor.
• S.E.: Standard error of the coefficient.
• Wald: Chi-square statistic for testing the significance of the coefficient.
• df: Degrees of freedom for the Wald test.
• Sig.: p-value associated with the Wald test.
• Exp(B): Odds ratio, representing the change in odds for a one-unit increase in
the predictor.
Interpretation:
• CGPA:
➢ Significance: Significant (p-value = 0.021)
➢ Effect: A one-unit increase in CGPA is associated with 410.208 times increase
in the odds of the outcome (positive effect).
• Specialization:
➢ Significance: Not significant for Specialization (1) or (2)
➢ Effect: The categories of Specialization do not have a significant impact on
the outcome.
• Aptitude Score:
➢ Significance: Significant (p-value = 0.094)
➢ Effect: A one-unit increase in Aptitude Score is associated with 1.665 times
increase in the odds of the outcome (positive effect).
 • Constant:
➢ Significance: Significant (p-value = 0.018)
➢ Effect: The constant term represents the log-odds of the outcome when all
predictors are zero.

Overall, the model suggests that CGPA and Aptitude Score are significant
predictors of the outcome. Specialization, as categorized, does not have a significant
effect.

5. What do the Betas in the Variables in Equation table signify? (Give

justification)

The Betas in the logistic regression model represent the log-odds of the outcome for
a one-unit increase in the corresponding predictor variable.
CGPA: (Beta 6.017)
Interpretation: For every one-unit increase in CGPA, the log-odds of the outcome
increase by 6.017. In other words, students with higher CGPA are more likely to
have a positive outcome (e.g., admission, success).
Specialization (1): (Beta 1.598)
Interpretation: Compared to the reference category (presumably Specialization 0),
students in Specialization 1 have a 1.598 higher log-odds of the outcome. This
suggests that students in Specialization 1 are more likely to have a positive outcome
than those in the reference category.
Specialization (2): (Beta: -0.317)
Interpretation: Compared to the reference category, students in Specialization 2
have a 0.317 lower log-odds of the outcome. This suggests that students in
Specialization 2 are less likely to have a positive outcome than those in the reference
category.
Aptitude Score: (Beta: 0.510)
Interpretation: For every one-unit increase in Aptitude Score, the log-odds of the
outcome increase by 0.510. This suggests that students with higher Aptitude Scores
are more likely to have a positive outcome.
Constant: (Beta: -62.249)
Interpretation: The constant term represents the log-odds of the outcome when all
predictors are zero. It provides a baseline for the model.