AI4ALL Module 3 Interactive Handout

AI4ALL · Stage 2

Workshop Learning Journey

A two-hour, fully hands-on module for government decision support.

2 hours

Theme: Understanding How Data Science and AI Support Better Government Decisions.

Main learning goal: By the end of this module, you should understand what data-driven decision making means; how AI can discover useful patterns and predict future outcomes; the difference between discovering patterns and making predictions; how Orange provides a visual, no-code way to perform AI analysis; and how decision makers can use AI results responsibly.

Understand the AI ProblemPrepare Data for AIPredict Future OutcomesDiscover Hidden PatternsSupport Better Decisions

Activity	Main question	Output
AI Learning Type	Does AI know the expected outcome, or is it discovering unknown patterns?	Classify organizational scenarios as supervised or unsupervised learning.
Data Preparation	Is the data ready and reliable enough for AI analysis?	A cleaned and prepared dataset.
Predict Future Outcomes	Can historical data help predict what is likely to happen next?	An attrition prediction model and responsible interpretation.
Discover Hidden Patterns	What groups can AI discover without knowing the answer in advance?	Service-performance clusters and possible KPI improvements.

Participant Details

Enter your details as you want them to appear in the completed PDF.

Full name *

Email address *

Department / organization *

Workshop date *

Activity 1

AI Learning Type

Classify practical examples and explain the clues behind your choices.

10–15 minutes

Objective: Understand the difference between Supervised Learning and Unsupervised Learning.

Background: Supervised Learning uses historical examples with known answers to predict future outcomes. Unsupervised Learning does not know the correct answers; it discovers hidden patterns, relationships, or natural groups.

Step 1: Classification Challenge

For each AI task, choose Supervised Learning or Unsupervised Learning. Do not worry if you are unsure—the discussion afterwards is the most important part.

#	AI Task	Classification

Step 2: Group Discussion

Why did you choose those answers? *

Which examples were difficult? *

What clues helped you decide? *

Activity 2

Data Preparation Process

Inspect, clean, prepare, and validate government service-request data.

30 minutes

Objective: Identify and resolve common data-quality problems before using data for AI and data analysis.

Preparation guidance: Check data quality for missing, duplicate, invalid, inconsistent, and unusual values; decide how to handle missing data; clean and standardize formats and categories; investigate outliers before removing them; select only relevant features; then transform and validate the final dataset.

InspectCleanHandle Missing DataSelect FeaturesTransformValidateAnalyze

Step 1: Inspect the Dataset

A government department collected the following service-request data for future AI analysis. Review the table and identify the problems.

Step 2: Prepare the Data

1. Check Data Quality

Missing values found *

Duplicate records found *

Invalid values found *

Inconsistent categories found *

Possible outliers found *

How many data-quality problems can you find? *

2. Handle Missing Data

Identify each missing value, choose how it should be handled, and explain why.

Record / missing attribute	Action	Reason

3. Clean & Standardize

Identify values that represent the same information but are written differently. Standardize the categories and address duplicates.

Original value or issue	Standardized value / action	Notes

4. Review Outliers

Decide which unusual values are clearly invalid and which could be genuine exceptional cases.

Record / unusual value	Assessment	Action and reason

5. Select Relevant Features

Assume the objective is to analyze patterns affecting service processing time and citizen satisfaction.

Attribute	Include or exclude?	Reason

6. Transform & Validate

Confirm each final preparation check.

Data types have been verified. Categorical and Yes/No attributes have been encoded if required. Numeric attributes have been scaled when appropriate. Missing and invalid values have been addressed. Categories are consistent. Duplicate records have been handled.

Final preparation notes *

Optional evidence: prepared data or Orange screenshot

Images are resized and saved only in this browser.

Activity 3

Predict Future Outcomes

Supervised Learning: build and interpret an employee-attrition prediction model.

40 minutes

Objective: Understand where AI can assist employees and where human involvement is still necessary.

Scenario: The HR department wants to identify employees at risk of resigning so managers can intervene early. Review the selected attributes before building the model in Orange.

Steps 1–2: Understand the Data and Identify Data Types

For every attribute, select the most appropriate data type. Data types may include unique ID, numeric, categorical, boolean, ordinal, textual, descriptive, date, time, media, or documental.

Attribute	Your selected data type

Discussion

Which attribute is the target for prediction in this scenario? *

Which attributes seem most influential? *

Which attributes do you consider unnecessary for prediction? *

Could we choose a different target for prediction? Give a scenario. *

Step 3: Build the Prediction Model Using AI

Follow the instructor in building the supervised learning system using Orange, then discuss:

Was the prediction correct? *

YesNoThe result was unclear

Should AI make the final decision? *

YesNoOnly with defined controls

When should a human review the application? *

Optional evidence: Orange workflow or prediction result

Upload a screenshot if requested by the instructor.

Activity 4

Discover Hidden Patterns

Unsupervised Learning: explore natural service-performance groups.

40 minutes

Objective: Understand how AI can discover hidden patterns that may not be obvious to humans.

Scenario: The Ministry wants to understand service performance before developing or reviewing KPIs. The AI does not know the groups in advance; it analyzes service data to discover patterns based on efficiency, quality, employee satisfaction, digital adoption, and resource utilization.

Steps 1–2: Understand the Data and Identify Data Types

Select the most appropriate data type for each service attribute. The middle column shows the values or scale used in the dataset.

Attribute	Values / scale in the dataset	Your selected data type

Discussion

Why is this dataset a good candidate for unsupervised learning? *

Which attributes do you consider unnecessary for clustering? *

Which attributes seem to be highly related? *

Which attributes represent service efficiency, service quality, and employee experience? *

Step 3: Discover Patterns Using AI

Build the unsupervised learning workflow in Orange. Use the suggested visualizations, then record what you observe.

1. Find services with low employee satisfaction and high complaint rates. What factors might explain this pattern?

Scatter PlotX: Employee_SatisfactionY: Complaint_RateColor: Cluster / Processing Time / Digital Usage / Workload / Cost

Your observation *

2. Find services with high employee workload and high cost per request.

Scatter PlotX: Employee_WorkloadY: Cost_Per_RequestColor: Cluster / Avg Processing Time

Your observation *

3. Do automated services appear in similar groups? How does their performance compare with non-automated services?

Scatter PlotX: Avg_Processing_TimeY: Complaint_RateColor: Automated_ProcessBox Plot: Processing Time / Complaint Rate

Your observation *

4. Do services with an existing KPI show different performance characteristics?

Box PlotSubgroups: KPI_Currently_DefinedSatisfaction / Complaints / Digital Usage / Cost / Workload

Your observation *

Analysis Summary

Optional evidence: Orange clustering workflow or results

Upload a screenshot if requested by the instructor.

Complete your handout

Final Reflection & Submission

Connect the workshop activities to your own department, then create and email your PDF.

Final step

0%

Your handout is in progress

Complete the remaining required fields before submission. You may still export a draft at any time.

Final Reflection

Download and email your completed handout

The page creates and downloads your PDF.
Your selected email application opens with a prepared message.
Attach the downloaded PDF, review the message, and select Send.

Send to *

Open email with