Moshraf Hossain

Let’s talk about data! Reach out via the links below.

Training Analytics Dashboard

Power BI Portfolio Project | Data-Driven Talent Development Insights

🎯 The Challenge

Training and development teams in organisations face a critical problem: they lack centralised, real-time visibility into training effectiveness and resource utilisation.

Imagine being an L&D manager responsible for hundreds of training sessions, multiple facilitators, and dozens of development projects—but all your data lives in disconnected spreadsheets. You spend 10-15 hours every month manually copying, pasting, and aggregating data just to answer basic questions:

How many employees are we reaching across different training programs?
Which facilitators are delivering the most sessions, and where are capacity gaps?
Are we creating training materials aligned with organisational priorities?
What’s our project completion rate, and where are the bottlenecks?

This was the business problem I set out to solve.

💡 My Solution

I built a comprehensive 3-page Power BI dashboard that transforms raw training data into executive-ready analytics. The solution provides role-specific insights for three key stakeholder groups, enabling data-driven decision-making without manual reporting overhead.

📊 Dashboard Architecture

The solution is structured as a 3-page analytical framework, with each page serving distinct stakeholder needs:

📄 Page 1: Executive Overview

Strategic KPIs and high-level trends for senior leadership

Executive Overview High-level KPIs and training activity trends for senior leadership decision-making

Key Metrics:

👥 Total Participants: 136 employees reached across all training sessions
📅 Sessions Delivered: 89 training activities completed
📚 Resources Created: 26 development projects tracked
✅ Project Completion Rate: 100% completion (demonstrating portfolio health)

Key Visualizations:

Training Activity Trend: Dual-axis chart revealing monthly patterns in session delivery vs. participant engagement
Resource Creation by Category: Donut chart showing content development priorities (Team Dev: 38%, Technical: 38%, Projects: 19%, Compliance: 4%)

Business Value: Executives can assess training impact in under 30 seconds without drowning in spreadsheet details. The dashboard immediately answers “Are we reaching our people?” and “Are we completing what we start?”

👥 Page 2: Training Delivery Analysis

Operational insights for training managers and coordinators

Training Delivery Analysis Facilitator performance metrics and session delivery insights for operational managers

Key Metrics:

📊 Total Sessions: 89 training activities delivered
👥 Total Participants: 136 individuals engaged
👨‍🏫 Active Facilitators: 4 trainers contributing to delivery
📈 Avg Participants/Session: 1.53 efficiency metric

Key Visualisations:

Top Facilitators by Sessions Delivered: Horizontal bar chart revealing Jenn leads with 37 sessions (41% of total delivery), while Roy delivered 14 sessions (16%)—highlighting potential workload imbalance
Sessions by Delivery Type: Donut chart showing 86.5% live vs. 13.5% virtual distribution
Monthly Session Delivery Trend: Area chart revealing peak activity in June-July (19 sessions each) and lower volume in October-November (5-7 sessions)—indicating seasonal patterns that inform hiring decisions

Business Value: Managers can identify workload imbalances (Jenn carrying 41% of delivery burden), understand delivery mode preferences, and anticipate seasonal capacity needs for proactive resource planning.

📂 Page 3: Resource Development Analysis

Project tracking and content creation insights for development teams

Resource Development Analysis Project tracking and resource creation metrics for content development teams

Key Metrics:

📂 Total Resources: 26 development projects in portfolio
✅ Projects Completed: 26 (100% completion rate)
📊 Completion Rate: 100% (calculated using robust DAX logic)
👤 Active Project Leads: 1 lead managing all projects

Key Visualisations:

Resources Created by Category: Column chart showing Team Dev and Technical lead with 10 resources each, Projects with 5, and Compliance with 1—revealing content distribution priorities
Project Status Distribution: Donut chart tracking Completed/In Progress/Not Started status (100% completed in this dataset)
Top Project Leads by Resources Created: Horizontal bar chart highlighting individual contributions
Resources by Complexity Level: 100% stacked bar chart revealing Team Dev projects are 90% Low complexity, while Technical projects split 50-50 between Low and Medium complexity

Business Value: Project leads can identify stalled initiatives, assess content complexity patterns, and align resource allocation with strategic priorities. The complexity breakdown helps with capacity planning—low complexity projects require less time investment.

🗂️ Data Model & Architecture

Star Schema Design

I designed a star schema optimised for analytical performance and maintainability:

Fact Tables:

Facts_Training_Session – Training delivery data (89 sessions, participant counts, delivery types)
Facts_Development_Projects – Resource creation tracking (59 projects, status, complexity levels)

Dimension Tables:

Dim_People – Facilitators and project leads (shared dimension with Role column)
Dim_Category – 6 training categories (Team Development, Technical, Projects, Compliance, Leadership, Onboarding)
Dim_Audience – 7 target audience groups
Dim_Date – Calendar table marked as date table for time intelligence functions

Relationship Structure:

All relationships are Many-to-One (*:1) for optimal query performance
Shared dimensions (Category, Audience) connect both fact tables via separate relationship paths
No many-to-many relationships—I resolved these by creating proper dimension tables

Why Star Schema?

Performance: Many-to-One relationships enable fast query execution even with large datasets
Maintainability: Shared dimensions reduce data duplication and simplify updates
Scalability: Easy to add new fact tables (e.g., participant feedback, course evaluations) without restructuring the entire model

Data Transformation Highlights

Power Query Steps:

Date Type Conversion
- Issue: Facts_Training_Session[Date] column stored as Text in source CSV
- Impact: Time-based filtering and chronological sorting not working
- Solution: Converted to proper Date data type in Power Query
- Result: Enabled time intelligence functions and date hierarchies
Dimension Table Creation
- Issue: Both fact tables contained Category and Audience columns, creating many-to-many relationship conflicts
- Solution: Extracted unique values to create separate Dim_Category and Dim_Audience tables
- Result: Clean star schema with proper Many-to-One relationships
Data Quality Validation
- Validated completeness of Status, Level, and Delivery Type columns
- Identified and handled null values appropriately
- Ensured consistent capitalization across categorical fields

🧮 DAX Measures Library

All measures are stored in a dedicated _Measures table for maintainability. Here are the key calculations:

Training Delivery Metrics

// Basic aggregation measure
Total Participants = SUM(Facts_Training_Session[Participants])

// Row count measure
Total Sessions = COUNTROWS(Facts_Training_Session)

// Efficiency metric with error handling
Avg Participants per Session = 
DIVIDE(
    [Total Participants],
    [Total Sessions],
    0
)
// The third parameter (0) prevents divide-by-zero errors when filters produce empty tables

// Distinct count for capacity tracking
Active Facilitators = DISTINCTCOUNT(Facts_Training_Session[Facilitator])

Resource Development Metrics

// Row count measure
Total Resources = COUNTROWS(Facts_Development_Projects)

// Filtered count using CALCULATE
Projects Completed = 
CALCULATE(
    COUNTROWS(Facts_Development_Projects),
    Facts_Development_Projects[Status] = "Completed"
)

// Percentage calculation with proper denominator
Completion Rate = 
DIVIDE(
    [Projects Completed],
    COUNTROWS(Facts_Development_Projects),
    0
)
// Formatted as Percentage with 0 decimal places in measure formatting

// Distinct count for team capacity
Active Project Leads = DISTINCTCOUNT(Facts_Development_Projects[Project Lead])

// Helper measure for chart aggregations
Project Count = COUNTROWS(Facts_Development_Projects)

Critical DAX Fix: Completion Rate

The Problem:

During development, I encountered a calculation error that produced inflated completion percentages:

Original Formula (Incorrect):

Completion Rate = DIVIDE([Projects Completed], [Total Resources], 0)

Issue: This formula divided completed projects by total resources—two conceptually different metrics. When filtered, this produced nonsensical results like 145% completion.

Root Cause Analysis:

[Projects Completed] counted projects with Status = “Completed”
[Total Resources] counted all rows in the development projects table
When both measures were filtered differently by slicers, the denominator didn’t match the numerator’s context

Corrected Formula:

Completion Rate = 
DIVIDE(
    [Projects Completed],
    COUNTROWS(Facts_Development_Projects),
    0
)

Why This Works:

Both numerator and denominator now use COUNTROWS() on the same table
They respond identically to filter context
The calculation correctly represents “completed projects ÷ total projects”

Result: Accurate 100% completion rate (26 completed ÷ 26 total) that reflects true portfolio health.

Lesson Learned: Always ensure that percentage calculations use compatible numerators and denominators that respond to filter context in the same way.

🎨 Design Principles & User Experience

Visual Design Standards

Colour Palette:

I selected colours with semantic meaning to guide user interpretation:

Primary Brand: #2E7D32 (Dark Green) – Headers, branding elements
Success Metrics: #4CAF50 (Green) – Positive outcomes, completed status
Neutral Metrics: #2196F3 (Blue) – General KPIs without positive/negative connotation
Progress Metrics: #FF9800 (Orange) – In-progress status, medium complexity
Calculated Metrics: #9C27B0 (Purple) – Derived calculations like averages and percentages

Status & Complexity Colours:

Completed: #4CAF50 (Green)
In Progress: #FF9800 (Orange)
Not Started: #E0E0E0 (Light Grey)
Low Complexity: #4CAF50 (Green)
Medium Complexity: #FF9800 (Orange)
High Complexity: #E53935 (Red)

Layout Standards

Technical Specifications:

Canvas: 1280px width, #F9F9F9 background for reduced eye strain
Header: 100px height, white text on green background
KPI Cards: 245px × 100px with 4px coloured left border for visual hierarchy
Charts: White background, 1px grey border, consistent 15px spacing
Typography: Segoe UI throughout with 9pt minimum for WCAG 2.1 AA accessibility compliance

UX Design Philosophy

1. Progressive Disclosure

Page 1 (Executive View): High-level summary answering “Are we effective?”
Pages 2-3 (Operational Views): Detailed drill-downs answering “Where specifically?”
Users start broad and drill down as needed—reducing cognitive overload

2. Consistent Navigation Pattern

Repeated header/filter/footer structure across all pages reduces learning curve
Users immediately know where to find controls
Footer shows “Page X of 3” for orientation

3. Semantic Colour Coding

Green always means success/completion
Orange always means progress/medium priority
Purple always means calculated/derived metrics
Reduces mental processing time—users decode meaning instantly

4. Accessible Contrast

All text meets WCAG 2.1 AA standards (minimum 4.5:1 contrast ratio)
KPI numbers use large fonts (32pt) for quick scanning
Chart data labels positioned for maximum readability

🛠️ Technical Challenges & Solutions

Challenge #1: Many-to-Many Relationship Error

Symptom: Power BI prevented direct relationship creation between fact tables and shared Category/Audience columns, displaying error: “This relationship has a many-to-many cardinality and can’t be used.”

Root Cause: Both Facts_Training_Session and Facts_Development_Projects contained Category and Audience columns. Power BI detected that a single category value appeared in multiple rows of both tables, creating a many-to-many relationship pattern.

Solution:

Created separate dimension tables:
- Dim_Category with unique category values
- Dim_Audience with unique audience values
Established Many-to-One relationships:
- Facts_Training_Session[Category] → Dim_Category[Category]
- Facts_Development_Projects[Category] → Dim_Category[Category]
- Similar pattern for Audience

Result: Clean star schema with proper relationships. Filters now cascade correctly from dimensions to both fact tables.

Learning: Always create dedicated dimension tables for shared attributes rather than reusing columns across fact tables. This is fundamental to star schema design.

Challenge #2: Completion Rate Calculation Error

Symptom: Completion Rate KPI displayed nonsensical values (e.g., 145%, or changing erratically with filters)

Root Cause: Original formula DIVIDE([Projects Completed], [Total Resources], 0) divided two conceptually different measures that responded differently to filter context.

Solution: Changed denominator to use COUNTROWS(Facts_Development_Projects) to ensure both numerator and denominator use the same table context.

Result: Accurate 100% completion rate that correctly reflects 26 completed projects out of 26 total.

Learning: Percentage calculations require compatible numerators and denominators. Both should reference the same table or respond identically to filters.

Challenge #3: Date-Based Filtering Not Working

Symptom:

Month names sorted alphabetically (April, August, December) instead of chronologically
Date slicers didn’t recognise the column as a date
Time intelligence functions threw errors

Root Cause: Source CSV stored dates as text strings (e.g., “2023-05-15” as Text data type)

Solution:

Opened Power Query Editor
Selected Facts_Training_Session[Date] column
Transform tab → Data Type → Date
Applied changes and refreshed data model

Result:

Proper chronological sorting in visuals
Date hierarchies (Year/Quarter/Month/Day) automatically available
Time intelligence functions (e.g., SAMEPERIODLASTYEAR) now functional

Learning: Always verify data types during initial data import. Text-formatted dates are a common issue in CSV files.

Challenge #4: Bar Chart Showing Identical Values

Symptom: “Top Project Leads” bar chart displayed the same value for all facilitators, despite different actual contribution levels.

Root Cause: Chart used the [Total Resources] measure instead of counting individual records per person. The measure returned the same grand total for each facilitator.

Solution: Changed the chart’s Values field from [Total Resources] to Count of Project ID, which correctly counts distinct projects per person.

Result: Accurate representation showing differential contributions (e.g., Lead A: 15 projects, Lead B: 8 projects).

Learning: When creating “per person” or “per category” charts, use row-level counts or sums rather than aggregate measures that might not respect the visual’s grouping context.

📈 Business Impact & Value Proposition

Quantifiable Outcomes

If deployed in a real organization, this dashboard would deliver:

Efficiency Gains:

✅ 80% reduction in reporting time: From 15 hours/month of manual Excel aggregation to <3 hours/month with automated refresh
✅ Instant insights: Stakeholders answer their own questions in seconds rather than waiting days for IT reports

Strategic Benefits:

✅ Self-service analytics: Executives, managers, and project leads access role-specific views without creating reporting bottlenecks
✅ Proactive decision-making: Seasonal patterns (June-July peaks) inform advance hiring for Q2 capacity
✅ Workload balancing: Identify facilitators carrying disproportionate loads (Jenn at 41% vs. Roy at 16%)

Resource Optimisation:

✅ Align content creation with demand: Category breakdown reveals whether resources match actual training delivery priorities
✅ Complexity-based planning: Understanding that Team Dev projects are 90% Low complexity helps estimate development timelines
✅ Completion tracking: 100% completion visibility prevents scope creep and ensures accountability

Stakeholder-Specific Value

For Executives:

Answer “Are we reaching our people?” at a glance (136 participants metric)
Assess portfolio health (100% completion rate)
Make budget decisions based on category distribution

For Training Managers:

Balance facilitator workloads proactively
Plan for seasonal capacity (hiring temps in May before June-July peak)
Understand live vs. virtual preferences (86.5% live—may need more physical training space)

For Project Leads:

Track project completion status across categories
Identify bottlenecks (all projects completed—process is effective)
Understand complexity patterns for future resource estimation

💼 Skills Demonstrated

Technical Capabilities

Power BI Platform:

✅ Report design and page layout optimization
✅ Interactive visualizations (9 chart types implemented)
✅ Slicer configuration for cross-page filtering
✅ Conditional formatting and semantic colour application
✅ Publishing and sharing workflows (prepared for Power BI Service deployment)

Data Modelling:

✅ Star schema design from first principles
✅ Relationship management (all Many-to-One, no many-to-many)
✅ Dimension table creation to resolve relationship conflicts
✅ Date table configuration for time intelligence
✅ Data model documentation and diagram creation

DAX (Data Analysis Expressions):

✅ Aggregation functions (SUM, COUNTROWS, DISTINCTCOUNT)
✅ Filter context manipulation with CALCULATE
✅ Error handling with DIVIDE and default return values
✅ Measure organization in dedicated tables
✅ Performance optimization (avoiding calculated columns where measures suffice)

Power Query (Data Transformation):

✅ Data type conversions (Text to Date)
✅ Column extraction and dimension table creation
✅ Data quality validation and null handling
✅ Source query optimization

Analytical & Business Skills

Business Analysis:

✅ KPI selection aligned to stakeholder needs (different metrics for each audience)
✅ Trend identification (seasonal patterns, workload distribution)
✅ Root cause analysis (why is Jenn carrying 41%? Capacity planning needed)
✅ Business requirements translation (L&D pain points → dashboard features)

Problem-Solving:

✅ Troubleshooting data model errors (many-to-many relationships)
✅ Debugging DAX calculation issues (completion rate formula)
✅ Iterative refinement (adjusted charts based on data availability)

Communication:

✅ Visual storytelling (progressive disclosure from high-level to detailed)
✅ Stakeholder-focused design (each page serves a specific role)
✅ Documentation and knowledge transfer (this portfolio page itself)

Design & User Experience

Visual Design:

✅ Colour theory application (semantic meaning for colours)
✅ Typography hierarchy (Segoe UI with consistent sizing)
✅ White space management for reduced cognitive load
✅ Consistent branding across all pages

Accessibility:

✅ WCAG 2.1 AA contrast compliance (4.5:1 minimum)
✅ Readable font sizes (9pt minimum)
✅ Colour-blind friendly palette (not relying solely on colour for meaning)

User Experience:

✅ Intuitive navigation (consistent header/filter/footer pattern)
✅ Self-explanatory visualizations (no manual required)
✅ Responsive to user interaction (slicers update all visuals instantly)

🚀 Future Enhancements

Predictive Analytics:

Implement DAX time intelligence functions to forecast training demand
Use historical patterns (June-July peaks) to predict future capacity needs
Alert managers 2 months in advance of high-demand periods

Drill-Through Capabilities:

Add participant-level drill-through pages (click a session → see attendee list)
Facilitator detail pages (click facilitator name → see full session history)
Session feedback integration (link to post-training survey scores)

Advanced DAX:

Create “same period last year” comparisons to track year-over-year growth
Implement running totals for cumulative participation tracking
Add “what-if” parameters for scenario planning (e.g., if we hire 2 more facilitators…)

Automated Alerts:

Use Power Automate to email managers when completion rate drops below 90%
Flag facilitators exceeding recommended session load (e.g., >35% of total)
Alert when project status hasn’t updated in 30+ days

Benchmarking:

Add industry benchmark data (average participants per session in similar organizations)
Compare performance across departments or regions
External data integration (LinkedIn Learning course completions, Coursera enrollments)

📂 Project Resources

GitHub Repository: View Full Documentation & Code

Project Files:

Training_Analytics_Dashboard_Portfolio.pbix – Power BI report file (interactive workbook)
data/Training_Delivery_Sessions.csv – Training session data (89 sessions, 2023)
data/Training_Development_Projects.csv – Resource development tracking (59 projects)
documentation/ – DAX measures, data model diagram, technical specifications

Download Dashboard: Training_Analytics_Dashboard_Portfolio.pbix

🎓 Reflections & Key Takeaways

What I Learned

Technical Lessons:

Data modelling is foundational – A clean star schema prevents countless downstream issues. Investing time upfront in proper dimension tables saved hours of troubleshooting later.
DAX requires careful validation – My completion rate error taught me to always test measures with multiple filter combinations. A formula that works with no filters might break when users apply slicers.
Data types matter – The date column issue (text vs. date) showed me that seemingly small data preparation steps have cascading impacts on visualization capabilities.

Design Lessons:

User experience trumps technical complexity – I could have added 20 more charts, but the three-page structure serves stakeholders better by reducing cognitive overload.
Semantic colour coding reduces mental load – Using green consistently for “success” and orange for “progress” means users decode meaning instantly without reading legends.

Process Lessons:

Documentation is essential – Detailed specifications enabled me to hand off this project confidently. In a real job, colleagues could maintain this dashboard without asking me constant questions.
Iterative development is realistic – I encountered 9 technical issues during development (documented in the handover report). Each issue taught me something new, and troubleshooting is where real learning happens.

Why This Project Matters for My Career

As a recent Master of Data Science graduate, this project demonstrates my readiness for Data Analyst roles by showcasing:

End-to-End Capability:

I didn’t just analyze existing dashboards—I built one from scratch, from data import to final visualization

Business Acumen:

I translated an ambiguous business problem (L&D teams need visibility) into specific, actionable features

Independent Problem-Solving:

I encountered and resolved 9 technical challenges without instructor guidance, demonstrating self-sufficiency

Stakeholder Awareness:

I designed three distinct pages because I understand that executives, managers, and project leads have different analytical needs

Communication Skills:

This portfolio page itself demonstrates my ability to explain technical work to non-technical audiences

Real-World Readiness:

The dashboard uses industry-standard tools (Power BI, DAX), follows best practices (star schema, semantic layer), and solves realistic business problems

📧 Let’s Connect

I’m actively seeking Data Analyst opportunities where I can apply Power BI, DAX, and business intelligence skills to drive data-informed decision-making.

Portfolio: https://moshraf.github.io/
LinkedIn: https://www.linkedin.com/in/moshrafhossain/
Email: mmoshraf@gmail.com
GitHub: https://github.com/moshraf

Interested in discussing this project or exploring collaboration opportunities? I’d love to connect!

Built with: Power BI Desktop | DAX | Star Schema Modelling | Power Query
Project Duration: 4 weeks (January - February 2026)
Status: Complete & Portfolio-Ready ✅

Tags: #PowerBI #DataAnalytics #DAX #BusinessIntelligence #DataModelling #PortfolioProject #DataScience

2026 1
2025 1
2023 5

2026