AI Predictions

Understanding Predictions

Our AI model analyzes historical data to predict:

  • Performance trends
  • Productivity patterns
  • Resource optimization opportunities

Manforce Dashboard User Guide

This user guide provides instructions for effectively navigating its features to extract valuable insights into Manforce worker performance (1 to 5 stars), feature weightage, and other key metrics. Scoring is done monthly with data collected since 2017.

1. Introduction to the Dashboard

This dashboard provides a visual and interactive interface for monitoring and analyzing Manforce worker-related data.

  • Login credential is required: Access dashboard using your credentials.
  • Time Period: Data spanning from January 1, 2017, to current date. Timeframe can be changed according to your preference (This/Last month/year).
  • Features for scoring: Retention, Disciplinary, Physical Fitness, Motivation, Prudence, Productivity, Compliance, Loyalty and Commitment.

2. Best Practices for Using the Dashboard

  • Start with the Overview Panel: Get a high-level understanding before diving into specific metrics.
  • Use Filters Effectively: Refine data views to gain targeted insights.
  • Reset Anytime: Clear all applied filters on a single chart (top left) or across the dashboard (top right).
  • Page Navigation: Navigate directly to relevant pages on the left column.
  • Multiple Charts Exploration: Some charts have drill-down options on the top right corner of the charts.
  • Sorting Preference: Sorting options are on the top right corner of the chart.

3. Main Sections of the Dashboard

  • Page 1: Overview Panel - Provides a quick executive summary of key data points and the demographic descriptions.
  • Page 2: Performance Insights with Overall Scoring - Scoring by all available data points. 1 scoring scheme for all data (feature weightages are the same for all customers).
  • Page 3: Performance Insights with Scoring by Customers - Scoring by customers (e.g., AEON BIG GROUP, PANASONIC). 1 scoring scheme for each customer (feature weightages might be different for different customers).
  • Page 4: Performance Insights with Scoring by Indicator Group - Scoring by indicator groups shared among the customers.
  • Page 5: Customer View - For future enhancement to let customers view their own profile only.
  • Page 6: Individual Worker Details - Select a single worker by using passport number or worker ID.

4. Scoring Approaches

Overall Scoring

  • Data Preparation: Worker data is loaded from dataset, cleaned, and normalized.
  • Score Calculation: Each worker gets a score based on their results across the indicators. The system looks at how their performance compares to everyone else overall and assigns a rating from 1★ (low) to 5★ (high).

Scoring by Company

  • Grouping: Data is grouped by customerName (company).
  • Score Calculation: Instead of comparing workers against all workers everywhere, the scoring here only compares workers within their own company.
  • Performance Label: Workers get a 1★–5★ rating, but only in the context of their own company.

Scoring by Indicator Group

  • Indicator Set Grouping: Workers are grouped by their “indicator set” (the set of features where they have nonzero values eg: Prudence, Motivation, Commitment).
  • Score Calculation: Workers are then compared with all workers that are in the same indicator group, which helps ensure that employees are only judged on the indicators that are related or relevant to them
  • Performance Label: Workers are assigned a performance label (1★ to 5★) within their indicator group.

Why Three Methods?

  • Overall Scoring: Compares all workers across the entire organization. This is useful for broad benchmarking — to see who the top and bottom performers are when everyone is looked at together.
  • By Company: Compares workers only within their own company. This adjusts for differences in company culture, policies, or work environments, making the scoring fairer for each company.
  • By Indicator Group: Compares workers who are measured using the same set of indicators (e.g., Motivation + Prudence, or Productivity + Compliance). This is helpful because not every company tracks the same things. By grouping companies with the same indicators, workers are only compared on the criteria that apply to them.

Logic

Performance depends on context. A score that looks “good” in one company might be average in another if expectations are different. By using these three methods, we make sure the scoring is:

  • Fair (workers are judged against the right peers),
  • Relevant (using the indicators that matter for them), and
  • Actionable (insights can be used at the organization, company, or group level).

5. Behind the Scoring System

Scoring Formula

For each worker, the score is calculated as follows:

  • For each feature (e.g., Motivation, Retention, Compliance, etc.):
    • Quantile thresholds ($q_1$, $q_2$, $q_3$) are computed from the data.
    • Weight ($w$) and polarity ($s$) are assigned to each feature.
    • The feature score is determined by comparing the worker’s value to the quantile thresholds:
      • If value $\geq q_3$: $4 \times w \times s$
      • If value $\geq q_2$: $3 \times w \times s$
      • If value $\geq q_1$: $2 \times w \times s$
      • If value $> 0$: $1 \times w \times s$
      • Else: $0$

Mathematically:

\[ \text{Total Score} = \sum_{\text{feature}} \text{Feature Score} \]

Indicators and How They Affect Scoring

Each worker is scored using several indicators (like Retention, Disciplinary, Motivation, etc.). Each indicator is made up of different data points and contributes differently to the final score. Some indicators are positive (the more, the better) while others are negative (the less, the better).

  • Retention: Looks at things like job changes, running away, short service, rejections, or terminations.
    • More of these events lowers the score (undesirable).
  • Disciplinary: Includes deductions, police reports, and counseling sessions.
    • Higher numbers mean lower scores (undesirable).
  • Physical Fitness: Includes medical leave days and medical expenses.
    • More health issues lower the score (undesirable).
  • Motivation: Looks at number of times allowances received and allowance pay amount.
    • Higher values increase the score (desirable).
  • Prudence: Based on savings count and total savings.
    • More savings increase the score (desirable).
  • Productivity: Measured using the onhold count (salary/payment holds).
    • Higher on-hold counts reduce the productivity score (undesirable).
  • Compliance: Tracks complaints (e.g., multiple types of compliance violations).
    • More complaints reduce the score (undesirable).
  • Commitment: Commitment is measured using overtime (OT) data.
    • Commitment reflects the worker's overtime (OT) level/amount. The higher the value the better the score.
  • Loyalty: Loyalty measured based on the worker's months of service.
    • Higher values improve the score (desirable).

6. Troubleshooting and Support

  • Data not loading: Refresh the page or check internet connectivity.
  • Filters not working: Clear filters and try reapplying them, alternatively, refresh the page.
  • Contact Support: For further assistance, submit a feedback form by navigating to the Dashboard-Feedback Section.

© 2025 Manforce Resources (M) Sdn Bhd. All rights reserved.