18  Predictive Analytics for HR Decision-Making

18.1 Overview of Predictive Dashboards

Predictive dashboards provide HR professionals with the ability to not only visualize current metrics but also forecast future trends and outcomes (Janet H. Marler & John W. Boudreau, 2017). By leveraging predictive analytics, HR can make informed decisions regarding workforce planning, retention strategies, and performance management.

18.1.1 Key Objectives

  • Understand the importance of predictive analytics in HR.
  • Learn how to integrate predictive models into dashboards.
  • Create actionable insights from predictive visualizations.

18.1.2 Introduction to Predictive Dashboards

Predictive dashboards combine historical data with advanced analytics to generate forecasts and trends. They enable organizations to:

  • Identify potential risks (e.g., high turnover rates).
  • Forecast recruitment needs based on company growth.
  • Predict employee performance trends over time.

18.1.3 Benefits of Predictive Dashboards in HR

  1. Proactive Decision-Making: Allows HR to anticipate challenges and act before they occur.
  2. Improved Resource Allocation: Helps in better planning for recruitment, training, and retention strategies.
  3. Enhanced Engagement: Forecasts engagement trends, enabling timely interventions.

18.1.4 Tools for Predictive Dashboards

  1. Power BI:
    • Features: Built-in forecasting tools, AI visualizations, integration with machine learning models.
    • Example: Forecasting attrition rates using historical employee data.
  2. Tableau:
    • Features: Advanced predictive analytics through integration with R and Python.
    • Example: Visualizing the likelihood of employee turnover by department.

18.2 Steps to Create Predictive Dashboards

  1. Data Preparation:
    • Gather historical HR data such as attrition, recruitment trends, and performance scores.
    • Clean and preprocess data for analysis.
  2. Model Building:
    • Use statistical or machine learning models to generate forecasts.
    • Common techniques include time series analysis, regression models, and classification.
  3. Visualization Design:
    • Import predictive results into Power BI or Tableau.
    • Use charts such as line graphs for trends or heatmaps for high-risk areas.
  4. Dashboard Interactivity:
    • Add slicers, filters, or drill-through features to enhance user experience.

Use Cases

  1. Turnover Prediction:
    • Predict future turnover rates to create targeted retention strategies.
  2. Recruitment Forecasting:
    • Forecast hiring needs based on organizational growth trends.
  3. Performance Prediction:
    • Identify employees likely to achieve or miss performance goals.

18.3 Hands-On Exercise: Designing a Predictive Dashboard for Employee Turnover in Power BI

18.3.1 Objective

Create a predictive dashboard in Power BI to forecast employee turnover rates based on historical data and visualize actionable insights.

Step 1: Dataset Overview

Dataset Details:
The dataset contains historical employee data with the following columns:

  1. EmployeeID: Unique identifier for each employee.
  2. Age: Employee age.
  3. Department: Department where the employee works.
  4. Tenure: Years of service.
  5. JobSatisfaction: Rating on a scale of 1 to 5.
  6. MonthlyIncome: Salary in dollars.
  7. PerformanceRating: Rating on a scale of 1 to 5.
  8. Overtime: Indicates if the employee works overtime (Yes/No).
  9. Attrition: Indicates if the employee left the company (Yes/No).

Step 2: Setting Up the Environment

  1. Open Power BI Desktop.
  2. Load the dataset by clicking on Home > Get Data > Excel (or CSV, depending on the file format).
  3. Preview and ensure the data types are correctly assigned (e.g., numeric, text).

Step 3: Data Preparation

  1. Clean Data:
    • Remove duplicates or irrelevant columns if present.
    • Ensure no missing values for key attributes like Attrition, Tenure, or PerformanceRating.
  2. Create New Columns:
    • Add a column for Tenure Group: Categorize employees into “0-2 years,” “3-5 years,” and “6+ years.”
    • Convert Overtime into binary values (Yes = 1, No = 0) for predictive modeling.
  3. Transform Attrition:
    • Convert Attrition into binary values (Yes = 1, No = 0) to use it as a target variable for prediction.

Step 4: Build Predictive Model in Power BI

  1. Go to Home > Transform Data to load Power Query.
  2. Use the AI Insights feature in Power BI to build a predictive model:
    • Select Attrition as the target variable.
    • Use predictors like Age, Department, Tenure, JobSatisfaction, Overtime, and MonthlyIncome.
    • Generate predictions and save them as a new column (e.g., Predicted Attrition).

Step 5: Create Visualizations

  1. Turnover Forecast Line Chart:
    • Add a line chart showing the predicted turnover rate by department over time.
    • X-Axis: Tenure Group.
    • Y-Axis: Predicted Attrition Rate (%).
  2. Attrition Risk Heatmap:
    • Create a heatmap to visualize attrition risk by Department and PerformanceRating.
    • Axis: Department (X-axis), Performance Rating (Y-axis).
    • Color Intensity: Predicted Attrition Rate.
  3. Departmental Summary Table:
    • Add a table visualization summarizing:
      • Department
      • Average Tenure
      • Predicted Attrition Rate
  4. Interactive Slicer:
    • Add slicers for filtering data by JobSatisfaction, Overtime, and Tenure Group.

Step 6: Interactivity and Design

  1. Add interactivity to the dashboard by linking slicers to all visuals.
  2. Use tooltips to show detailed employee information when hovering over a data point.
  3. Format visuals for clarity:
    • Use appropriate labels, titles, and data colors.
    • Highlight departments with the highest turnover risk.

Step 7: Insights and Action Items

Analyze the dashboard to derive insights:
1. Identify departments with high predicted attrition rates.
2. Observe the impact of overtime on attrition trends.
3. Propose actionable strategies, such as:

  • Revising overtime policies.
  • Offering incentives to employees with low job satisfaction.

Discussion

  1. What are the key factors contributing to employee turnover in this dataset?
  2. How can HR intervene based on the predictive insights?
  3. What additional data might improve the model’s accuracy?

18.4 Hands-On Exercise: Designing a Predictive Dashboard for Employee Turnover in Tableau

18.4.1 Objective

Create a predictive dashboard in Tableau to forecast employee turnover rates and visualize actionable insights.

Step 1: Dataset Overview

Dataset Details:
The dataset should contain the following columns:

  1. EmployeeID: Unique identifier for each employee.
  2. Age: Employee age.
  3. Department: Department where the employee works.
  4. Tenure: Years of service.
  5. JobSatisfaction: Rating on a scale of 1 to 5.
  6. MonthlyIncome: Salary in dollars.
  7. PerformanceRating: Rating on a scale of 1 to 5.
  8. Overtime: Indicates if the employee works overtime (Yes/No).
  9. Attrition: Indicates if the employee left the company (Yes/No).

Step 2: Load Dataset

  1. Open Tableau Desktop.
  2. Click Connect to Data and select your data source (Excel, CSV, or database).
  3. Preview and verify data types. Ensure numerical columns (e.g., Tenure, MonthlyIncome) are correctly identified.

Step 3: Data Preparation

  1. Create Calculated Fields:
    • Convert Attrition into binary (Yes = 1, No = 0):
Code
IF [Attrition] = "Yes" THEN 1 ELSE 0 END
  • Create a Tenure Group field:
Code
IF [Tenure] <= 2 THEN "0-2 years"
ELSEIF [Tenure] <= 5 THEN "3-5 years"
ELSE "6+ years"
END
  • Calculate Attrition Risk (%):
Code
SUM([Attrition Binary]) / COUNT([Attrition Binary]) * 100
  1. Aggregate Data:
    Group data by Department, Tenure Group, and other key metrics for analysis.

Step 4: Build Visualizations in Tableau

  1. Turnover Forecast Line Chart:
    • Rows: Predicted Attrition Risk (%) (calculated field).
    • Columns: Tenure Group.
    • Add Department as a color dimension to compare trends.
  2. Attrition Risk Heatmap:
    • Rows: Department.
    • Columns: Performance Rating.
    • Marks: Set to Square and use Predicted Attrition Risk (%) as the color intensity.
  3. Departmental Summary Table:
    • Use a text table to display:
      • Department
      • Average Tenure
      • Predicted Attrition Risk (%)
  4. Interactive Filters:
    • Add filters for JobSatisfaction, Overtime, and Tenure Group using the Show Filter option.
    • Place filters on the side of the dashboard for user interaction.

Step 5: Add Predictive Analytics

  1. Use Tableau’s Forecasting Tool:
    • Right-click on the line chart and select Add Forecast.
    • Choose an appropriate model (e.g., exponential smoothing) and adjust the forecast length.
  2. Integrate External Predictive Models (optional):
    • If more advanced predictions are required, integrate a model created in R or Python using Tableau’s R Integration or TabPy.

Step 6: Create the Dashboard

  1. Go to Dashboard > New Dashboard.
  2. Drag and drop the visualizations created earlier onto the dashboard.
  3. Add interactivity:
    • Use Actions to link filters and drill-downs across visuals.
  4. Format the dashboard:
    • Use consistent color schemes and add titles for clarity.
    • Highlight departments with high turnover risk using alerts or annotations.

Step 7: Insights and Interpretation

Analyze the dashboard to derive insights:
1. Identify departments with the highest attrition risk.
2. Observe how overtime impacts predicted attrition.
3. Discuss actionable strategies to reduce attrition.

Discussion

  • What patterns do you observe in turnover trends?
  • How does predictive modeling improve HR decision-making?
  • What additional data could enhance the accuracy of predictions?