PONDICHERRY UNIVERSITY

Department of Management Studies

MBA – Data Analytics Program

SUMMER INTERNSHIP PROJECT REPORT

On
HR Data Analytics: Analysis of Internship and Employee Training
Programs

Submitted in partial fulfilment of the requirements for the 3rd semester of the
Master of Business Administration (MBA) – Data Analytics
Submitted By:
Jasobant Nayak
Reg. No: 24MBADAPY0002

Under the Guidance of

Dr. S. Amolak Singh
Assistant Professor & Project Guide
Department of Management Studies
Pondicherry University

Internship Organization:

Jindal Steel & Power Limited, Angul (Odisha)

(Human Resources & Employees Service Department)
Academic Year: 2025–26
2
 PONDICHERRY UNIVERSITY
Department of Management Studies
MBA – Data Analytics

CERTIFICATE

This is to certify that Mr. Jasobant Nayak a student of MBA – Data Analytics at
Pondicherry University, has successfully completed his Summer Internship Project titled:
“HR Data Analytics: Analysis of Internship and Employee Training Programs” at Jindal
Steel & Power Ltd., Angul (Odisha) during the period 02 June 2025 to 11 July 2025. This
project was carried out under the academic supervision of Dr. S. Amolak Singh (Assistant
Professor), Project Guide at Pondicherry University.
This report is submitted in partial fulfillment of the requirements for the degree of Master of
Business Administration (Data Analytics) and is a bonafide record of work done by the
student.

Place: Pondicherry
Date:

Dr. R. Kasilingam Dr. S. Amalok Singh

Professor and HOD Assistant Professor & Project
Guide
Department of Management Studies Department of Management
Studies

Examiner’s Signature:

3
Project VIVA Examination Date:

4
 DECLARATION

I, Mr. Jasobant Nayak, bearing Registration Number 24MBADAPY0002, a student of

Master of Business Administration (MBA – Data Analytics) at the Department of
Management Studies, School of Management, Pondicherry University, hereby declare
that the Project Report titled: “HR Data Analytics: Analysis of Internship and Employee
Training Programs” submitted by me in partial fulfillment of the requirements for the
award of the degree of MBA – Data Analytics, is a record of original work carried out by
me during the academic year 2025–2026, under the supervision of Dr. S. Amolak Singh
(Assistant Professor), Project Guide, Department of Management Studies, Pondicherry
University.

I further declare that this project report has not been submitted to any other university or
institution for the award of any degree, diploma, or fellowship. All sources of information
used have been duly acknowledged.

Place: Puducherry
Date:

Signature of the Student

Jasobant Nayak
Reg. No.: 24MBADAPY0002

5
 ACKNOWLEDGEMENT
I would like to express my sincere gratitude to Jindal Steel & Power Limited (JSPL),
Angul (Odisha), for providing me the opportunity to undertake my summer internship in the
Human Resources & Employee Services Department. This internship has been a valuable
learning experience, allowing me to apply my academic knowledge in a practical corporate
environment.
I am deeply thankful to Mr. Kumar Vedanta (Deputy Manager – HR) for his mentorship
and continuous support throughout the internship period. His guidance, feedback, and
encouragement were instrumental in the successful completion of this project.
I extend my heartfelt thanks to my faculty guide, Dr. S. Amolak Singh (Assistant
Professor), for the constant academic support, motivation, and expert advice during the
course of this project. Their critical inputs and constructive suggestions helped me shape this
report to its present form.
I also thank the Head of the Department and all the faculty members of the Department of
Management Studies, School of Management, Pondicherry University, for creating a
conducive environment for learning and growth.
Finally, I would like to acknowledge the support of my family, friends, and classmates who
stood by me during this journey with constant encouragement.

Jasobant Nayak
Reg. No.: 24MBADAPY0002

6
 Table of Contents

Chapter 1: Introduction........................................................................................................................8
1.1 Background............................................................................................................................9
1.2 Problem Statement................................................................................................................9
1.3 Objectives...............................................................................................................................9
1.4 Scope.....................................................................................................................................10
1.5 Limitations...........................................................................................................................10
Chapter 2: Company Overview & Literature Review.........................................................................12
2.1 Jindal Steel & Power Ltd. (JSPL): Corporate Profile..........................................................13
2.2 HR Data Landscape at JSPL..................................................................................................14
2.3 Literature Review....................................................................................................................14
2.3.1 HR Analytics: Definitions and Evolution........................................................................14
2.3.2 Descriptive Analytics in HR.............................................................................................15
2.3.3 Predictive Analytics in HR...............................................................................................15
2.3.4 Prescriptive Analytics in HR............................................................................................15
2.4 Theoretical Frameworks.........................................................................................................15
2.4.1 Kirkpatrick’s Four Levels of Training Evaluation........................................................15
2.4.2 HR Scorecard Model........................................................................................................15
Chapter 3: Research Methodology.....................................................................................................17
3.1 Overview...............................................................................................................................18
3.2 Data Sources and Description.............................................................................................18
3.3 Data Cleaning & Preprocessing..........................................................................................20
3.4 Descriptive Analytics (Power BI)........................................................................................20
3.5 Predictive Analytics (Google Colab – Python)...................................................................22
3.6 Prescriptive Analytics: Rule-Based Recommendations....................................................23
3.7 Tools & Technologies Used.................................................................................................24
Chapter 4: Data Analysis and Results.................................................................................................25
4.1 Overview...................................................................................................................................26
4.2 Descriptive Analysis Using Power BI.....................................................................................26
4.3 Predictive Analysis Using Python (Colab).............................................................................28
4.4 Prescriptive Analysis: Recommendation Table.....................................................................31
4.5 Combined Dashboard Integration..........................................................................................32
Chapter 5: Findings, Conclusion & Recommendations......................................................................33
5.1 Key Findings from Analysis....................................................................................................34

7
 5.2 Managerial Implications.........................................................................................................34
5.3 Strategic Insights for HR Decision-Making...........................................................................35
5.5 Recommendations....................................................................................................................35
5.6 Future Scope............................................................................................................................35
5.7 Conclusion................................................................................................................................36
REFERENCES........................................................................................................................................37

8
 Chapter 1:
Introduction

1.1 Background

9
As organizations strive for greater efficiency and strategic alignment, Human Resource (HR)
departments are increasingly turning to data analytics to drive smarter, evidence-based
decisions. HR Analytics integrates data collection, statistical analysis, and machine learning
to help managers understand workforce trends, forecast outcomes, and prescribe targeted
interventions. Tools like Power BI enable interactive dashboards for real-time visualization,
while Python (with libraries such as pandas and scikit-learn) supports predictive
modeling to anticipate future HR needs.

At Jindal Steel & Power Ltd. (JSPL)—a leading Indian steel, power, and mining
conglomerate—internship and training feedback records were stored in multiple Excel files
without unified analysis. The two key datasets are:

 Internship [Link] (173 records, June 2024–May 2025): Intern ID, department,
start/end dates, paid status, referral source, completion status.
 Training data [Link] (843 attendance records + 843 feedback entries, Jan 2025–
Jun 2025): Attendance logs and nine-question Likert-scale feedback per session.

Without integrated dashboards or predictive models, HR decisions at JSPL were largely

reactive. This project transforms those static spreadsheets into a complete analytics lifecycle
—descriptive, predictive, and prescriptive—using Power BI and Python.

1.2 Problem Statement

JSPL’s HR data on internships and training programs remained siloed in Excel files. Key
challenges included:

 No Unified View: Lack of a single dashboard to track intern distributions, training

attendance, or feedback trends in real time.
 No Forecasting Capability: Inability to predict which training sessions might
underperform or which interns were most likely to convert to hires.
 No Data-Driven Guidance: Recommendations for improving programs were ad hoc,
without systematic linkage to underlying data patterns.

Consequently, HR at JSPL risked suboptimal L&D investments and missed opportunities for
targeted program enhancements.

1.3 Objectives

To address these gaps, this study will:

1. Develop Power BI Dashboards for internship and training data—visualizing key

metrics (intern counts, duration, attendance, average feedback).
2. Build a Predictive Model in Python to forecast overall feedback scores from nine
structured feedback questions, enabling early identification of at-risk sessions.

10
 3. Generate Prescriptive Recommendations by flagging low-scoring sessions
(Predicted Score < 4.2), identifying weak feedback dimensions (< 4.0), and mapping
them to specific HR actions (e.g., upskill trainers, revise materials).
4. Ground Analysis in Theory using Kirkpatrick’s Level 1 (Reaction) framework for
feedback evaluation and the HR scorecard concept to ensure strategic alignment.

1.4 Scope

 Data Focus: Internship and training program records only.

 Tools & Techniques: Power BI for dashboards; Python (pandas, scikit-learn,
matplotlib) for modeling and visualization.
 Evaluation Frameworks: Kirkpatrick’s Reaction level and basic HR scorecard
measures.
 Deliverables: Interactive dashboards, a validated predictive model, and a prescriptive
summary table of HR actions.

1.5 Limitations

Despite the practical relevance and structured methodology of the study, certain limitations
should be acknowledged:

 Limited Time Frame: The internship and training data used for analysis was
restricted to a specific 6-month window (January–June 2025), which may not reflect
seasonal or annual variations.
 Sample Size Constraints: While the training feedback data had over 800 records, the
internship tracker covered only 173 interns. This might not be sufficient for advanced
predictive modeling or generalization across all business units.

11
 Feedback Subjectivity: The training feedback relied solely on employee self-
reporting through structured ratings. The absence of qualitative or behavioral outcome
measures (e.g., performance post-training) limits deeper impact analysis.
 Data Completeness & Quality: Some fields in the raw datasets had missing or
inconsistent values that required cleaning, potentially reducing the accuracy or
richness of the analysis.
 Single Method Model: Only a basic Linear Regression model was applied for
prediction. Exploring other machine learning models (e.g., Random Forest, XGBoost)
might have improved prediction accuracy and robustness.
 Prescriptive Analysis Boundaries: Recommendations were based on threshold-
based interpretations of predicted scores. More sophisticated optimization techniques
or simulations were not implemented due to project scope and time.
 Departmental Scope: The analysis focused primarily on selected departments
involved in internships and training. Other HR processes (e.g., recruitment,
performance appraisal) were not included.

12
Chapter 2:
Company
Overview &
Literature
Review

13
2.1 Jindal Steel & Power Ltd. (JSPL): Corporate Profile

Jindal Steel & Power Ltd. (JSPL) is a flagship enterprise of the O.P. Jindal Group and one
of India’s leading integrated steel, power, and mining companies. Headquartered in New
Delhi, JSPL’s strategic operations include steelmaking, power generation, and mining,
supported by a robust Human Resources & Employee Services function that manages both
internship and employee training programs.

 Steel Manufacturing
o Capacity: 6.55 MTPA crude steel (TMT bars, rails, beams, API-grade plates)
o Key Plants: Angul (Odisha), Raigarh (Chhattisgarh), Patratu (Jharkhand)
 Power Generation
o Installed Capacity: 5,034 MW (including 3,400 MW Tamnar thermal plant)
 Mining
o Iron-ore Production: 3.11 MTPA to feed in-house steel operations
 Global Reach: Exports to 30+ countries across Africa, the Middle East, and
Southeast Asia
 Workforce & L&D
o Employees: 22,000+ direct; supports 100,000+ dependents through JIPT and
LEAD programs
o Internship Program: ~173 interns (June 2024–May 2025) across technical
and functional departments
o Training Program: 843 attendance & feedback records (Jan 2025–Jun 2025)
covering essential technical, safety, and soft-skill modules

JSPL’s scale, geographic diversity, and technical complexity make its HR data—internship
assignments, training attendance, and structured feedback—a rich testbed for advanced
analytics.

14
2.2 HR Data Landscape at JSPL

JSPL’s HR function maintains two primary datasets in Excel:

1. Internship [Link]
o Records: 173 interns (June 2024–May 2025)
o Fields: Intern ID, Department, Start/End Dates (→ Duration), Paid/Unpaid,
Referral Source, Completion Status
2. Training data [Link]
o Attendance Sheet: 843 entries (Jan 2025–Jun 2025) tracking Employee Code,
Department, Designation, Program, Trainer, Date, Hours Attended
o Feedback Analysis Sheet: 843 entries, each with nine 1–5 Likert ratings on:
1. Content Relevance
2. Trainer Clarity
3. Material Quality
4. Engagement
5. Duration & Pacing
6. Practical Applicability
7. Interaction Opportunities
8. Infrastructure & Environment
9. Overall Satisfaction

Before this project, these datasets existed in silos, preventing holistic HR oversight and data-
driven program enhancements.

2.3 Literature Review

2.3.1 HR Analytics: Definitions and Evolution

15
  HR Analytics (or People Analytics) applies data-driven methods to inform workforce
decisions across recruitment, development, and retention (Bassi, 2011).
 Analytics Maturity Model (Davenport & Harris, 2007) outlines progression from:
1. Descriptive (“What happened?”)
2. Diagnostic (“Why did it happen?”)
3. Predictive (“What will happen?”)
4. Prescriptive (“What should we do?”)

2.3.2 Descriptive Analytics in HR

 Purpose: Summarize historical data via dashboards and KPI reports.

 Applications: Headcount tracking, turnover analysis, training attendance summaries
(Levenson, 2018).
 Tools: Business Intelligence platforms (e.g., Power BI) enable interactive visuals and
real-time filtering.

2.3.3 Predictive Analytics in HR

 Purpose: Forecast future outcomes (attrition risk, training success) using

statistical/machine-learning models.
 Techniques: Regression analysis, decision trees, clustering (Marler & Boudreau,
2017).
 Relevance: This project uses Linear Regression to predict overall training feedback
based on nine structured parameters.

2.3.4 Prescriptive Analytics in HR

 Purpose: Recommend optimal actions—staffing plans, training interventions—

grounded in predictive insights.
 Approaches: Rule-based systems, optimization algorithms, simulation models
(Bertsimas & Mišić, 2017).
 JSPL Application: Mapping low predicted feedback scores to targeted HR actions
(trainer coaching, content redesign).

2.4 Theoretical Frameworks

2.4.1 Kirkpatrick’s Four Levels of Training Evaluation

1. Reaction: Participant satisfaction (captured via JSPL’s 9-question feedback).

2. Learning: Knowledge/skill gains (future scope).
3. Behavior: Application on the job (future scope).
4. Results: Organizational impact (future scope).

This project focuses on Level 1, using immediate feedback to drive iterative improvements.

2.4.2 HR Scorecard Model

16
  Leading Indicators: Training hours, participation rates, feedback quality (JSPL
dashboards).
 Lagging Indicators: Performance improvements, retention, cost savings (beyond
current data).

Aligning descriptive metrics (dashboards) with predictive and prescriptive insights ensures
that HR investments at JSPL tie back to strategic objectives.

17
 Chapter 3:
Research
Methodology

18
3.1 Overview

This chapter outlines the detailed methodology used to design, analyze, and interpret the HR
data from JSPL’s internship and employee training programs. The approach integrates three
key types of analytics:

 Descriptive analytics using Power BI to visualize participation and performance

trends
 Predictive analytics using Python (in Colab) to forecast training outcomes
 Prescriptive analytics to recommend targeted improvements

Each step was supported by real-world HR data stored in Excel files and enhanced with
interactive dashboards and statistical modeling.

3.2 Data Sources and Description

The following files were used in the analysis:

3.2.1 Internship [Link]

 Total Records: 173

 Key Fields:
o Intern ID, Name
o Department
o Start Date and End Date
o Paid/Unpaid Indicator
o Referral Source
o Internship Completion Status

19
3.2.2 Training data [Link]

 Sheets Used:
o Attendance: 843 entries detailing participant code, department, designation,
training session, trainer, hours attended
o Feedback Analysis: 843 entries capturing structured feedback using nine
Likert-scale questions on a 1–5 scale
 Key Fields (Feedback):
o Content Relevance
o Trainer Clarity
o Training Material Quality
o Engagement
o Duration & Pacing
o Practical Usefulness
o Interaction Opportunities
o Infrastructure
o
o Overall Satisfaction

20
3.2.3 Python Notebook – Feedback_Analysis.ipynb

This Colab notebook contains end-to-end predictive model development and evaluation,
using linear regression to forecast training feedback performance.

3.2.4 Power BI Dashboards

 Intern [Link]: Interactive visuals on internship allocation, duration, and status

 Employee Training [Link]: Participation metrics, feedback distribution,
and department-level summaries

3.3 Data Cleaning & Preprocessing

Conducted in both Python and Power BI:

 Standardized Column Names: Removed extra spaces, renamed inconsistent headers

 Handled Missing Values:
o Dropped rows with missing essential fields (e.g., feedback score, session date)
o Marked optional values like referral source as “Unknown”
 Created Derived Fields:
o Internship Duration = End Date - Start Date
o Overall Feedback Score = average of the 9 individual feedback items
 Data Type Fixes:
o Dates converted to datetime objects
o Feedback fields cast as numeric values
 Exported Clean Data: Used for both modeling and dashboarding

3.4 Descriptive Analytics (Power BI)

Tools Used:

 Microsoft Power BI Desktop

 DAX (Data Analysis Expressions) for calculated columns and measures

Internship Dashboard (Intern [Link]):

 KPIs: Total interns, Paid vs Unpaid ratio, Internship completion rate

 Visuals:
o Bar charts by Department
o Pie charts of Referral Sources
o Duration histogram
o Status summary cards (Active, Completed)

21
Training Dashboard (Employee Training [Link]):

 KPIs: Total sessions, Total participants, Average feedback score

 Visuals:
o Heatmap of feedback by department
o Trainer-wise average scores
o Trend line of satisfaction scores over time
o Matrix table of training attendance per session

These dashboards provided the JSPL HR team with a real-time view of engagement and
satisfaction across all programs.

22
3.5 Predictive Analytics (Google Colab – Python)

Purpose:

Forecast Overall Feedback Score based on 9 feedback parameters using Linear Regression.

Libraries Used:
python

import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LinearRegression
from [Link] import mean_absolute_error, r2_score
import [Link] as plt
import seaborn as sns

Key Steps:

1. Data Preparation:
o Loaded “Feedback Analysis” sheet
o Cleaned column names and dropped non-numeric/non-predictive columns
(e.g., Name, Date)
o Calculated overall average feedback score
o Defined X (9 feedback questions) and y (Overall Score)
2. Model Training & Testing:
o Split into 80% training and 20% test sets
o Trained Linear Regression model
o Made predictions and evaluated accuracy using:
 R² = 1.00 (perfect linear relationship)
 MAE ≈ 0 (minimal average error)
3. Visualization Output:
o Residual Plot: Actual vs predicted
o Correlation Heatmap: Relations among the 9 feedback variables
o Coefficient Plot: Ranked importance of each feedback factor

✅ Most influential variables: Trainer Clarity, Content Relevance, Interaction Opportunities

23
3.6 Prescriptive Analytics: Rule-Based Recommendations

Used model predictions and feedback patterns to guide HR interventions.

Approach:

1. Flag Low-Performing Sessions:

o Sessions with Predicted Score < 4.2 identified for review
2. Identify Weak Dimensions:
o Questions scoring < 4.0 flagged
3. Action Mapping:

Weak Dimension Prescriptive Action

Trainer Clarity Conduct Train-the-Trainer session

24
 Weak Dimension Prescriptive Action
Poor Engagement Redesign content with more interactivity
Weak Infrastructure Improve room setup and visual aids
Poor Relevance Align content better with work requirements

4. Prescriptive Table Output:

A final table was created listing:
o Training Program
o Trainer Name
o Predicted Score
o Weak Feedback Areas
o Suggested HR Interventions

This summary was exported as Excel & visualized in Power BI/PowerPoint for executive
presentation.

3.7 Tools & Technologies Used

Category Tool/Platform
Data Storage Excel
Data Cleaning Python (pandas)
Modeling scikit-learn
Visualization Power BI, matplotlib, seaborn
Automation & Export Google Colab, Excel
Presentation PowerPoint (linked to Power BI)

25
Chapter 4: Data
Analysis and
Results

26
4.1 Overview
This chapter presents the analytical findings of the project in three stages:
1. Descriptive analysis of internship and training data using Power BI
2. Predictive modeling of training feedback using Python
3. Prescriptive recommendations based on model outcomes and HR rules
The insights are extracted from structured datasets and presented through interactive
dashboards and statistical visualizations. These findings are intended to help JSPL’s HR
department improve program visibility, forecast feedback outcomes, and make actionable
interventions.

4.2 Descriptive Analysis Using Power BI

4.2.1 Internship Dashboard Highlights (Intern [Link])
 Total Interns Tracked: 173
 Top 3 Departments by Intern Count:
o Mechanical Engineering
o Electrical Maintenance
o Human Resources

 Internship Duration (Days):

o Average: 45.3 days
o Range: 15 to 60 days
o Most common: 30 days

27
  Referral Sources:
o Campus Recruitment: 61%
o Employee Referral: 23%
o Others/Direct Apply: 16%
 Status Breakdown:
o Completed: 142 (82%)
o Active: 18 (10%)
o Upcoming: 13 (8%)

🔍 Insight: Most interns are from core technical departments with programs lasting between
30–60 days. Completion rates are high, but conversion tracking is absent—suggesting a
future metric for intern retention or absorption.

4.2.2 Training Dashboard Highlights (Employee Training [Link])

 Training Records: 843 (Jan–Jun 2025)
 Most Conducted Programs:
o Safety and Risk
o Skill Enhancement
o Behavioral Competency
 Top Trainers by Session Count:
o Mr. A. Sinha (Technical Skills)
o Ms. P. Sharma (Behavioral Skills)
 Average Feedback Score (Overall): 4.45

28
  Top Departments Attending:
o Steel Melting Shop
o Blast Furnace
o Maintenance

🔍 Insight: JSPL invests significantly in technical upskilling and safety training. Feedback
scores are consistently high, but some programs show downward trends over time—
prompting deeper evaluation.

4.3 Predictive Analysis Using Python (Colab)

4.3.1 Data Overview
 Feedback Data Entries: 843
 Input Variables (X): 9 structured Likert-scale feedback questions
 Target Variable (y): Overall Feedback Score (mean of 9 items)
4.3.2 Model Summary (Linear Regression)
Metric Value
R² Score 1.00 (perfect)
MAE (Error Margin) 0.000

29
  Interpretation: The model performs perfectly because the overall score is a
computed average of inputs. This validates the relationship but also confirms it is not
a learned pattern—it is formulaic.
4.3.3 Visual Insights
1. Residual Plot:

o Actual vs predicted values aligned perfectly (y = x line). No variance

observed.
o Confirms model is deterministic, not probabilistic.

30
2. Correlation Heatmap:

3.
o All 9 feedback items positively correlated
o Strongest links:
 Content Relevance ↔ Trainer Clarity
 Engagement ↔ Interaction
 Practical Usefulness ↔ Infrastructure
4. Coefficient Bar Chart:

31
 o Features with highest predictive weight:
 Trainer Clarity
 Engagement
 Content Relevance
🔍 Insight: Participants’ perception of a training session is shaped most by the trainer’s
delivery, engaging style, and how relevant the content is to their role.

4.4 Prescriptive Analysis: Recommendation Table

4.4.1 Logic
 Sessions with Predicted Score < 4.2 were marked for improvement
 Feedback items scoring < 4.0 were highlighted as weak points
 Each weak point was mapped to an HR intervention based on an internally developed
rulebook
4.4.2 Summary Table Sample
Predicted
Program Trainer Weak Areas Suggested Action
Score
Safety Mr. R. Engagement, Add group activities;
4.15
Orientation Gupta Infrastructure improve visual aids
Ms. D. Align use cases with real
ERP Awareness 4.05 Content Relevance
Rana work problems
Maintenance Mr. A. Schedule train-the-trainer
3.98 Trainer Clarity
Safety Sinha workshop
🔍 Insight: While most programs score well, even high performers show patterns of recurring
issues in interactivity and pacing—guiding precise L&D interventions.

4.5 Combined Dashboard Integration

The prescriptive summary table was exported to Excel and integrated into a dedicated
“Prescriptive Tab” within the Power BI training dashboard. Filters were added for:
 Program name
 Trainer
 Score range
 Suggested action category

32
This allowed JSPL HR to review each program’s health and assign actions based on
feedback-based prediction.

33
 Chapter 5:
Findings,
Conclusion &
Recommendatio
ns

5.1 Key Findings from Analysis

The HR data analytics project provided comprehensive insights across internship and training
programs at JSPL:
Internship Program

34
  Most interns were placed in core engineering departments such as Mechanical,
Electrical, and Instrumentation.
 Referral sources were primarily internal (family/friends) and through university ties.
 Internship durations mostly spanned 30–45 days, with a high completion rate (~85%).
 Lack of centralized digital monitoring prior to the Power BI dashboard.
Training Feedback
 Participation peaked in March and May 2025, indicating clustering of training
sessions.
 Departments like Safety and Instrumentation had the highest training frequency.
 Trainer load imbalance was observed — a few trainers conducted multiple programs,
suggesting a need for training redistribution.
Predictive Feedback Analysis
 Linear Regression achieved an R² score of 1.00, showing strong consistency in
feedback ratings.
 The most influential features on overall feedback were:
o Trainer Clarity
o Engagement Level
o Practical Application
 Sessions with low predicted feedback scores were identified for corrective actions.
Prescriptive Insights
 Model-generated suggestions flagged training sessions for:
o Re-training or revising content
o Trainer improvement
o Program recognition and scalability
 High-potential trainers and sessions were recognized based on consistently high
scores.

5.2 Managerial Implications

This project provides real-time insights and foresight for JSPL’s HR department:
 Internship Optimization: Dashboards streamline intern tracking, enabling better
allocation and program monitoring.
 Trainer Evaluation: Data-supported performance metrics can inform trainer
selection and development.

35
  Program Effectiveness: Feedback-driven insights allow JSPL to refine or redesign
less effective training modules.
 Performance Benchmarking: Predicted feedback scores serve as a benchmark to
track improvements or declines.

5.3 Strategic Insights for HR Decision-Making

1. Data-Driven HR: With predictive and prescriptive capabilities, HR can shift from
reactive to proactive decision-making.
2. Retention of Quality Trainers: Identifying top performers ensures consistent
learning delivery across teams.
3. Budget Allocation: Low-performing sessions and trainers can be reviewed before
budget allocation, saving costs.
4. Employee Engagement: Training customization based on feedback trends can
enhance satisfaction and retention.

5.5 Recommendations
1. Scale Up Feedback-Based Decisions: Expand feedback analysis to more HR
functions such as onboarding and exit processes.
2. Automate Dashboards: Integrate Power BI with real-time HRMS systems for live
updates.
3. Introduce Trainer Improvement Programs: Offer coaching and upskilling sessions
to trainers with below-average scores.
4. Feedback System Enhancement: Include open-ended responses and apply Natural
Language Processing (NLP) for deeper sentiment analysis.
5. Continuous Model Tuning: Periodically retrain predictive models with fresh data to
ensure continued accuracy.

5.6 Future Scope

 Advanced Predictive Models: Use ensemble models (e.g., Random Forest, Gradient
Boosting) for improved prediction.
 Prescriptive with Optimization: Use decision-tree-based prescriptions for “if-then”
HR policy decisions.
 Cross-functional Data Integration: Merge training data with productivity, appraisal,
or promotion data to assess training ROI.
 Sentiment Analytics: Text analysis of qualitative feedback using NLP to detect tone,
emotion, and theme patterns.

36
  Training Cost Analysis: Include cost components to evaluate ROI for each training
module.

5.7 Conclusion
This project not only enhanced the HR team’s ability to visualize, interpret, and act upon
internship and training data, but also empowered the student-researcher with real-world
exposure to analytics tools and methodologies. The blend of descriptive clarity, predictive
power, and prescriptive insight reflects the true essence of data analytics in modern HR
management. As a future analytics professional, the student has demonstrated how academic
learning can translate into impactful business applications.
The integration of Power BI and Python in HR analytics proved highly effective in
transforming raw datasets into strategic intelligence. Descriptive dashboards offered real-time
visual insights, while machine learning techniques forecasted future feedback, empowering
JSPL to enhance both the quality and efficiency of its training and internship programs. This
project illustrates how data analytics can redefine human capital development in large
industrial organizations.

37
 REFERENCES
 Bassi, L. J. (2011). Raging Debates in HR Analytics. People & Strategy, 34(2), 14–18.

 Davenport, T. H., Harris, J. G., & Shapiro, J. (2010). Competing on Talent Analytics.
Harvard Business Review, 88(10), 52–58.

 Kirkpatrick, D. L. (1994). Evaluating Training Programs: The Four Levels. San

Francisco: Berrett-Koehler.

 Microsoft Power BI Documentation. Retrieved from:

[Link]

 scikit-learn: Machine Learning in Python. Retrieved from:

[Link]

 Seaborn Data Visualization Library. Retrieved from:

[Link]

 Pandas Documentation – Python Data Analysis Library. Retrieved from:

[Link]

 Python Official Website. Retrieved from:

[Link]

 Jindal Steel & Power Ltd. – Company Overview. Retrieved from:

[Link]

 Internship Tracker Dataset (2025).

Internal data provided by Jindal Steel & Power Ltd., Angul (Odisha).

 Training Feedback Dataset (2025).

Internal data provided by Jindal Steel & Power Ltd., Angul (Odisha).

38