Analytics
ProBeya's analytics engine provides statistical analysis and visualization tools for operational performance data. It transforms raw KPI values and board data into actionable insights through trend analysis, Pareto charts, cross-board comparisons, and anomaly detection.
Trend Analysis
Trend analysis examines a single KPI's performance over time, providing statistical context and highlighting deviations from expected behavior.
What Trend Analysis Includes
For each KPI, trend analysis computes:
| Metric | Description |
|---|---|
| Raw values | Time-series data points with dates |
| Moving average | Smoothed trend line (configurable window size, default: 5 points) |
| Mean | Arithmetic average of all values in the time range |
| Standard deviation | Measure of value dispersion around the mean |
| Percentiles | P10, P50 (median), and P90 for distribution context |
| Trendline | Linear regression line showing the overall direction |
| Anomalies | Data points that deviate more than 2 standard deviations from the mean |
Using Trend Analysis
- Open a KPI detail panel by clicking on a KPI card.
- Navigate to the Trend tab.
- Select a time range from the presets or set custom dates.
- The trend chart displays with the moving average overlay and anomaly markers.
Time Range Presets
| Preset | Duration | Best For |
|---|---|---|
| 7d | Last 7 days | Daily KPIs — recent performance snapshot |
| 30d | Last 30 days | Weekly KPIs — month-over-month comparison |
| 90d | Last 90 days | Quarterly review — seasonal pattern identification |
| 1y | Last 12 months | Annual trends — long-term trajectory |
| Custom | User-defined start and end dates | Specific analysis periods |
Moving Average Window
The moving average smooths out short-term fluctuations to reveal the underlying trend. The window size determines how many data points are averaged:
- Window = 3 — Responsive to changes, more noise visible
- Window = 5 — Balanced smoothing (default)
- Window = 7 — Heavy smoothing, best for noisy data
- Window = 10-30 — Very heavy smoothing, shows only macro trends
Anomaly Detection
Anomalies are data points that fall outside 2 standard deviations from the mean. They are highlighted with markers on the trend chart and listed separately for review.
Anomalies may indicate:
- Equipment failure — Sudden drop in OEE or yield
- Process change — Step change in defect rate after a new material batch
- Data entry error — Unusually high or low value that should be verified
- One-time event — Planned shutdown, holiday effect, or special batch
Click on an anomaly marker to view the KPI value's comment field, which often contains an explanation for the unusual value (e.g., "Machine 3 down for maintenance").
Multi-KPI Comparison
Compare 2-5 KPIs on the same time axis to identify correlations and trade-offs between metrics.
How It Works
- Navigate to the analytics view from the board toolbar.
- Select Multi-KPI Comparison.
- Choose 2-5 KPIs to compare.
- Set the date range for the comparison.
The chart displays all selected KPIs on a shared time axis. Since KPIs may use different units (%, ppm, count), values are normalized to a 0-100 scale for visual comparison.
Normalization
Each KPI's values are normalized using min-max scaling within the selected time range:
normalized = (value - min) / (max - min) * 100
This allows KPIs with different scales (e.g., OEE at 85% and defect rate at 120 ppm) to be meaningfully compared on the same chart.
The original values and units are preserved in tooltips and the data table below the chart.
Statistical Summary
For each KPI in the comparison, the following statistics are displayed:
| Statistic | Description |
|---|---|
| Mean | Average value across the time range |
| Std Dev | Standard deviation (variability measure) |
| Min | Lowest value recorded |
| Max | Highest value recorded |
Pareto Analysis
Pareto analysis identifies the "vital few" causes that account for the majority of a problem — the classic 80/20 rule.
Creating a Pareto Chart
- Open a board and navigate to the analytics view.
- Select Pareto Analysis.
- Choose the column to analyze (e.g., "Deviation Type", "Root Cause Category").
- Optionally apply filters:
- Time range — Restrict to items created within a specific window
- Category column — Filter by a secondary column (e.g., show Pareto for deviation types in Department X only)
- Team filter — Filter by assignee (proxy for team/department)
- Click Generate.
Pareto Chart Components
The Pareto chart displays:
- Bar chart — Categories ranked by frequency (highest first)
- Cumulative line — Running cumulative percentage overlaid on the bars
- 80% threshold line — Horizontal line marking the 80% threshold
- Vital few count — Number of categories that account for 80% of occurrences
Interpreting Results
Category | Frequency | % | Cumulative %
─────────────────┼───────────┼───────┼─────────────
Equipment fault | 45 | 36.0% | 36.0%
Material defect | 30 | 24.0% | 60.0%
Operator error | 20 | 16.0% | 76.0%
Process drift | 12 | 9.6% | 85.6% ← 80% threshold
Calibration | 8 | 6.4% | 92.0%
Environmental | 5 | 4.0% | 96.0%
Other | 5 | 4.0% | 100.0%
In this example, the top 4 categories (equipment fault, material defect, operator error, process drift) account for 85.6% of all deviations. These are the "vital few" that should receive focused improvement effort.
Pareto analysis results are bounded to 10,000 items per query to prevent performance issues on very large boards. If your board exceeds this limit, apply time range filters to narrow the dataset.
Cross-Board KPI Comparison
Compare the same KPI across multiple boards within the organization. This enables benchmarking between production lines, departments, or sites.
How It Works
- Navigate to the workspace-level analytics view.
- Select Cross-Board Comparison.
- Enter the KPI name to compare (case-insensitive match).
- Optionally restrict to specific boards.
- Set the time range for the comparison.
Comparison Output
For each board that has a matching KPI:
| Field | Description |
|---|---|
| Board name | The board containing this KPI |
| Workspace | The workspace the board belongs to |
| Latest value | Most recent KPI value |
| Mean | Average value over the time range |
| Standard deviation | Variability measure |
| Trend | Linear regression slope (improving/declining/stable) |
| Status | Current traffic-light status (green/amber/red) |
Performance Ranking
Boards are ranked by their latest KPI value, taking the KPI direction into account:
- Higher is better KPIs (e.g., OEE): ranked highest value first
- Lower is better KPIs (e.g., defect rate): ranked lowest value first
Organization Benchmark
The comparison also calculates an organization-wide benchmark — the average of the latest values across all boards. This provides a reference point for identifying top performers and underperformers.
Board | Latest OEE | vs. Benchmark (86.2%)
─────────────────────┼────────────┼──────────────────────
Production Line 1 | 92.1% | +5.9% ↑
Production Line 3 | 88.5% | +2.3% ↑
Production Line 2 | 84.7% | -1.5% ↓
Packaging Line | 79.4% | -6.8% ↓
Permissions
| Action | Required Role |
|---|---|
| View trend analysis | Any board member |
| View Pareto charts | Any board member |
| View cross-board comparison | Any workspace member |
| Export analytics data | Board admin or workspace admin |
Related Features
- KPI Boards — KPI definitions and values are the primary data source for analytics
- Action Log — Pareto analysis helps prioritize which issues to create actions for
- Reports & Export — Analytics data can be included in scheduled PDF reports
- Formula Engine — Computed KPIs feed into analytics the same way as manual KPIs