Based on “A Dean's Dilemma…” - HBR case
Name: Sudheer Yerra
Roll No: 2025EPGP063
Case: Dean's Dilemma: Selection of Students for the MBA Program
Authors: Dhimant Ganatra, Dinesh Kumar Unnikrishnan
Dataset: Shared over email on May 12th
Date: 06/06/2025
Objective
• Purpose: To identify suitable candidates for MBA admission using historical performance indicators.
• Dataset: 391 candidates with features like entrance test score, academic scores, work experience,
placement outcome…etc.
• Columns used:
• Percentile_ET
• Percent_Degree
• Experience_Yrs
• Placement
Admission Rule
• Since the objective is to admit highly suitable candidates, we need de ne a set of conditions which
will lter out bad candidates
• These are the 4 conditions I de ned:
1. Percentile_ET >= 75
2. Percent_Degree >= 70
3. Experience_Yrs >= 1
4. Placement= Placed
• This is the r script I used:
• data$Recommended_Admits <- ifelse(data$Percentile_ET >= 75 & data$Percent_Degree >= 70 &
data$Experience_Yrs >= 1 & data$Placement == "Placed", "Yes", "No")
fi
fi
fi
Data Visualizations
T-Test
Goal: Check di erences in Entrance test percentile (Percentile_ET) and Placement status (Placement)
Null Hypothesis (H₀): There is no di erence between Percentile_ET & Placement
Alternate Hypothesis (H₁): There is a di erence between Percentile_ET & Placement
These are the values our test gave:
• t= -2.5991 (indicates stronger di erence) | df= 120.22
• p-value= 0.01052
• Since the value is < 0.05, we reject the null hypothesis
• Mean of Placed= 56.99; Mean of Not Placed= 46.83
• It means that students with Percentile_ET will be mostly placed
ff
ff
ff
ff
Pearson Correlation Test
We got the following metrics from the test:
1. Correlation coe cient (r)= 0.213
This shows that there is weak positive correlation between the two columns
2. p-value= 2.14e-05
Since the p-value is < 0.05 we reject H₀
It means co ee consumption has no signi cant impact on productivity
3. Con dence interval= (0.116, 0.306)
With 95% con dence we can say that true correlation lies in this range
fi
ff
fi
ffi
fi
Linear Regression Test
1. Regression slope= 0.745
For every 1% in Percent_Degree, Percentile_ET increases by 0.745
2. Intercept= 8.0103
If Percent_Degree= 0, Percentile_ET = 6.4333
3. F-statistic= 18.51; p-value= 2.14e-05
Both indicate that the overall model is not signi cant
4. R-squared= 0.04542
Only 4.5% of the variation in Percentile_ET is explained by Percent_Degree marks.
This is very low, indicating a poor model t.
fi
fi
Conclusion
• Once all these tests are done, we are able to nd the students who are quali ed to get the admit.
• Out of all the 391 students, only 15 students met the admission criteria
• The rest of the 376 candidates have to be rejected
fi
fi