Sairis

Appearance

Data Analytics - Comprehensive User Guide ​

Overview ​

The Data Analytics tool provides natural language access to your structured data stored in datasets. It works with ANY dataset - sales, HR, inventory, support tickets, or custom data - without requiring SQL knowledge or manual configuration.

Key Capabilities ​

✅ Domain Agnostic - Works with any dataset and column names
✅ Natural Language - Ask questions in plain English
✅ Financial Accuracy - All calculations verified with MathTool (50-digit precision)
✅ Smart Matching - Fuzzy column and value matching
✅ Rich Analysis - Grouping, ranking, trending, forecasting, and more

What Makes This Different ​

  • No SQL Required - Natural language queries work automatically
  • Zero Configuration - Works with any dataset structure immediately
  • Financial Grade - Precision calculations with audit trails
  • Intelligent - Understands partial names, fuzzy values, context

Getting Started ​

Basic Query Structure ​

Simply ask questions about your data:

"show me records"
"list all datasets"
"what columns are available?"
"group by status"

Query with Context (Advanced) ​

Add filters using the | separator:

"sum amount by salesperson | status=closed won"
"show top 10 deals | exclude status closed lost"
"count records | date=2024-01-01 to 2024-12-31"

Core Features ​

1. Dataset Discovery ​

List Available Datasets ​

What it does: Shows all datasets you can query

Examples:

"list datasets"
"show all datasets"
"what datasets are available?"
"show me all available data"

Sample Response:

markdown
## Available Datasets

**Total Datasets:** 3

### 1. sales_opportunities
- **Records:** 1,247
- **Columns:** 15
- **Last Updated:** 2024-10-20

### 2. employee_roster
- **Records:** 543
- **Columns:** 22
- **Last Updated:** 2024-10-15

### 3. inventory_items
- **Records:** 8,921
- **Columns:** 18
- **Last Updated:** 2024-10-21

2. Column Inspection ​

List Columns in Dataset ​

What it does: Shows all column names and data types

Examples:

"list columns"
"show columns"
"what fields are available?"
"describe dataset structure"
"show top 10 columns"

Sample Response:

markdown
## Dataset Schema: sales_opportunities

**Total Columns:** 15
**Total Records:** 1,247

### Column List:
1. **Opportunity Name** (text) - 1,247 non-null
2. **Sales Person** (text) - 1,247 non-null
3. **Deal Amount** (float64) - 1,195 non-null (95.8% complete)
4. **Status** (text) - 1,247 non-null
5. **Close Date** (datetime) - 1,180 non-null (94.6% complete)
6. **Lead Source** (text) - 1,247 non-null
7. **Product Category** (text) - 1,230 non-null (98.6% complete)
...

Show Column Schema with Samples ​

What it does: Shows detailed column information with sample values

Examples:

"show dataset schema"
"describe columns with examples"
"what data types are in the dataset?"

Sample Response:

markdown
## Detailed Column Schema

**Sales Person** (text)
- Non-null: 1,247 (100%)
- Unique Values: 45
- Sample Values: "Jordan Faas", "Sarah Miller", "John Smith"

**Deal Amount** (float64)
- Non-null: 1,195 (95.8%)
- Unique Values: 892
- Sample Values: "$125,000", "$87,500", "$250,000"
- Statistics: Min: $5,000 | Max: $2,500,000 | Avg: $156,432

**Status** (text)
- Non-null: 1,247 (100%)
- Unique Values: 7
- Sample Values: "Closed Won", "Lead", "Qualified", "Closed Lost"

3. Grouping & Aggregation ​

Group by Single Column ​

What it does: Counts or sums records by category

Examples:

"group by status"
"count by sales person"
"total by product category"
"breakdown by region"
"sum amount by status"

Sample Response:

markdown
## Grouping Analysis: Total by Status

**Total Records:** 1,247
**Number of Groups:** 7

### Breakdown by Group:

#### Closed Won
- **Count:** 342 records (27.4%)
- **Total Deal Amount:** $52,450,000
- **Average Deal Amount:** $153,362

#### Lead
- **Count:** 485 records (38.9%)
- **Total Deal Amount:** $48,500,000
- **Average Deal Amount:** $100,000

#### Qualified
- **Count:** 280 records (22.5%)
- **Total Deal Amount:** $35,000,000
- **Average Deal Amount:** $125,000

**Grand Total:** $135,950,000
**Math Tool Verified:** ✅

Group by Multiple Columns (Pivot) ​

Examples:

"show sales by region and product"
"pivot amount by salesperson and status"
"breakdown by quarter and status"

4. Sorting & Ranking ​

Sort Records ​

What it does: Orders records by column values

Examples:

"sort by deal amount descending"
"show records ordered by close date"
"list by sales person ascending"
"rank by total amount"

Sample Response:

markdown
## Dataset: sales_opportunities

**Sorted by:** Deal Amount (descending)
**Total Rows:** 1,247 | **Showing:** 50 rows

| Sales Person | Deal Amount | Status | Close Date |
|--------------|-------------|--------|------------|
| Jordan Faas | $2,500,000 | Closed Won | 2024-09-15 |
| Sarah Miller | $1,850,000 | Closed Won | 2024-08-22 |
| John Smith | $1,200,000 | Qualified | 2024-11-30 |
...

Top/Bottom N ​

Examples:

"show top 10 deals by amount"
"bottom 5 salespeople by revenue"
"highest 20 products by quantity"
"lowest 10 inventory items by value"

5. Filtering ​

Basic Filtering ​

What it does: Shows records matching criteria

Examples:

"show records where status = closed won"
"filter by sales person = jordan faas"
"show only active employees"
"exclude status closed lost"

Query Format with Filters:

"sum amount | status=closed won"
"group by salesperson | status=closed won, exclude status=lead"
"show records | date=2024-01-01 to 2024-12-31"
"count by category | status=active, region=west"

Comparison Filtering ​

Examples:

"show deals over $100,000"
"products with quantity below 10"
"employees with less than 2 years tenure"
"deals between $50,000 and $200,000"

Exclude Filters ​

Examples:

"group by status | exclude status=closed lost"
"sum amount | exclude salesperson=jane doe"
"show records | exclude category=inactive"

6. Statistical Analysis ​

Summary Statistics ​

What it does: Shows min, max, mean, median, standard deviation

Examples:

"summarize deal amounts"
"show distribution of salaries"
"statistics for product prices"
"what are the quartiles for deal sizes?"

Sample Response:

markdown
## Statistical Summary: Deal Amount

**Total Records:** 1,247
**Valid Values:** 1,195 (95.8%)

| Statistic | Value |
|-----------|-------|
| Count | 1,195 |
| Mean | $156,432 |
| Median | $125,000 |
| Std Dev | $87,234 |
| Min | $5,000 |
| 25th Percentile | $85,000 |
| 50th Percentile | $125,000 |
| 75th Percentile | $200,000 |
| Max | $2,500,000 |

**Outliers Detected:** 23 records (1.9%)
**Math Tool Verified:** ✅

7. Percentages & Ratios ​

Percentage Breakdown ​

What it does: Calculates percentage of total for each group

Examples:

"what percent of deals are closed won?"
"show percentage breakdown by status"
"calculate win rate by salesperson"
"what's the distribution by category?"

Sample Response:

markdown
## Percentage Breakdown by Status

**Total Records:** 1,247

| Status | Count | Percentage |
|--------|-------|------------|
| Lead | 485 | 38.9% |
| Qualified | 280 | 22.5% |
| Closed Won | 342 | 27.4% |
| Closed Lost | 95 | 7.6% |
| On Hold | 30 | 2.4% |
| Disqualified | 15 | 1.2% |

**Math Tool Verified:** ✅ Total: 100.0%

Win Rate / Conversion Rate ​

Examples:

"calculate win rate by salesperson"
"what's the conversion rate from lead to closed?"
"show success rate by product"
"completion rate by team"

8. Calculated Columns ​

Create New Columns with Formulas ​

What it does: Performs calculations across columns

Examples:

"calculate commission = deal amount * 0.15"
"compute total = quantity * price"
"add column for profit margin = (revenue - cost) / revenue"
"show daily rate = deal amount / days"

Sample Response:

markdown
## Dataset with Calculated Column: Commission

**Formula:** Commission = Deal Amount × 0.15
**Total Rows:** 1,247 | **Showing:** 50 rows

| Sales Person | Deal Amount | Commission |
|--------------|-------------|------------|
| Jordan Faas | $500,000 | $75,000.00 |
| Sarah Miller | $350,000 | $52,500.00 |
| John Smith | $250,000 | $37,500.00 |
...

**Total Commission:** $20,392,500.00
**Math Tool Verified:** ✅

Advanced Features ​

9. Fuzzy Value Matching ​

What it does: Matches partial names/values without exact spelling

How it works: When you mention a partial value, the system finds matching records

Examples:

"group by jordan"              → Matches "Jordan Faas"
"sum amount by faas"           → Matches "Jordan Faas"
"group by engineering"         → Matches "Engineering" department
"show records for smith"       → Finds all "Smith" employees
"total by tech"                → Matches "Tech Support" team

Sample Response:

markdown
## Fuzzy Match Found

**Search Term:** "jordan"
**Matched Value:** "Jordan Faas"
**Column:** Sales Person
**Records Found:** 87

### Breakdown by Status:
- **Closed Won:** 24 deals ($3,450,000)
- **Qualified:** 31 deals ($3,875,000)
- **Lead:** 32 deals ($3,200,000)

**Total:** 87 deals | **$10,525,000**

Multiple Match Handling:

If "smith" matches multiple people:

markdown
## Multiple Matches Found

**Search Term:** "smith"
**Matches Found:** 3 in Sales Person column

Please be more specific:
1. John Smith (45 records)
2. Jane Smith (38 records)
3. Bob Smith (22 records)

Try: "group by john smith" or "group by jane smith"

10. Temporal Analysis & Trending ​

What it does: Analyzes how your data changes over time - works with ANY dataset that has dates

Time Periods Supported: ​

  • Hourly - Hour-by-hour analysis
  • Daily - Day-by-day trends
  • Weekly - Week-by-week patterns
  • Monthly - Month-over-month trends
  • Quarterly - Quarter-by-quarter analysis (Q1, Q2, Q3, Q4)
  • Yearly - Year-over-year comparisons

Domain-Agnostic Examples: ​

ANY Dataset:

"show monthly trend"
"breakdown by quarter"
"weekly analysis"
"yearly comparison"
"temporal breakdown by month"

Sales Data:

"monthly revenue trend"
"quarterly deal count"
"sales by month"

HR/Employee Data:

"hiring trend by quarter"
"turnover rate by month"
"headcount by year"

Inventory Data:

"stock levels by week"
"inventory value trend monthly"
"product turnover quarterly"

Support Tickets:

"ticket volume by week"
"resolution time trend monthly"
"escalations by quarter"

Safety/TRIFR Data:

"TRIFR by quarter"
"injury rate trend"
"safety metrics by month"
"recordable incidents quarterly"

Sample Response:

markdown
## Temporal Analysis: Monthly Breakdown

**Time Period:** Monthly
**Date Range:** Jan 2024 - Oct 2024
**Total Records:** 1,247

| Period | Record Count | Total Value | Avg Value | Unique Entities | Change MoM |
|--------|-------------|-------------|-----------|-----------------|------------|
| Jan 2024 | 95 | $12,450,000 | $131,053 | 42 | - |
| Feb 2024 | 102 | $13,875,000 | $136,029 | 45 | +11.4% ↑ |
| Mar 2024 | 118 | $15,200,000 | $128,814 | 48 | +9.5% ↑ |
| Apr 2024 | 89 | $11,650,000 | $130,899 | 40 | -23.4% ↓ |
| May 2024 | 125 | $16,800,000 | $134,400 | 52 | +44.2% ↑ |
...

**Total:** $135,950,000 across 10 months
**Average per Period:** 124.7 records | $13,595,000
**Trend:** Growing (+8.5% average per month)
**Math Tool Verified:** ✅

**Auto-Detected Columns:**
- Date: "Close Date"
- Value: "Deal Amount"
- Entity: "Sales Person"

Period-over-Period Comparisons ​

Month-over-Month (MoM):

"show MoM growth"
"month over month change"
"compare this month to last month"

Quarter-over-Quarter (QoQ):

"compare Q1 vs Q2"
"show quarterly growth"
"QoQ change"
"quarterly comparison"

Year-over-Year (YoY):

"show year over year change"
"compare 2023 vs 2024"
"YoY growth"
"annual comparison"

Sample QoQ Response:

markdown
## Quarter-over-Quarter Analysis

| Quarter | Total Value | Change QoQ | Growth Rate |
|---------|-------------|------------|-------------|
| Q1 2024 | $41,525,000 | - | - |
| Q2 2024 | $47,350,000 | +$5,825,000 | +14.0% ↑ |
| Q3 2024 | $46,175,000 | -$1,175,000 | -2.5% ↓ |

**Best Quarter:** Q2 2024 ($47.35M)
**Growth Trend:** +5.6% average
**Math Tool Verified:** ✅

11. Trending Analysis with Growth Rates ​

What it does: Identifies trends, growth patterns, and momentum in your data

Automatic Detection:

  • Trend direction (growing, declining, stable)
  • Average growth rate per period
  • Growth acceleration/deceleration
  • Seasonal patterns
  • Anomaly detection

Domain-Agnostic Examples:

"show trend"
"what's the growth rate?"
"trending analysis"
"identify trend direction"
"growth pattern"

Sample Response:

markdown
## Trending Analysis: Monthly Pattern

**Analysis Period:** Jan 2024 - Oct 2024 (10 months)
**Trend Direction:** **Growing** ↑

### Growth Metrics:
- **Average Growth Rate:** +8.5% per month
- **Total Growth:** +127% (Jan to Oct)
- **Acceleration:** Increasing momentum
- **Volatility:** Moderate (±12% variation)

### Monthly Growth Rates:
| Month | Value | Growth | Rate |
|-------|-------|--------|------|
| Jan | $12.45M | - | - |
| Feb | $13.88M | +$1.43M | +11.4% |
| Mar | $15.20M | +$1.32M | +9.5% |
| Apr | $11.65M | -$3.55M | -23.4% |
| May | $16.80M | +$5.15M | +44.2% |
...

### Insights:
- **Peak Month:** May 2024 ($16.80M)
- **Lowest Month:** April 2024 ($11.65M)
- **Strongest Growth:** May 2024 (+44.2%)
- **Trend Confidence:** High (consistent pattern)

**Math Tool Verified:** ✅

12. Forecasting & Predictions ​

What it does: Projects future values based on historical trends

Forecasting Methods:

  • Linear trend projection
  • Moving average forecasting
  • Seasonal adjustment
  • Confidence intervals

Domain-Agnostic Examples:

"forecast next 3 months"
"predict next quarter"
"project next 6 periods"
"extrapolate next year"
"forecast future trend"

With Time Periods:

"forecast next 3 months monthly"
"predict next 4 quarters quarterly"
"project next year yearly"

Sample Response:

markdown
## Forecast Analysis: Next 3 Months

**Historical Data:** Jan - Oct 2024 (10 months)
**Forecast Method:** Linear regression with seasonal adjustment
**Confidence Level:** 95%

### Forecast Results:

| Period | Forecasted Value | Lower Bound | Upper Bound | Confidence |
|--------|-----------------|-------------|-------------|------------|
| Nov 2024 | $17,250,000 | $15,120,000 | $19,380,000 | 95% |
| Dec 2024 | $18,100,000 | $15,540,000 | $20,660,000 | 95% |
| Jan 2025 | $18,800,000 | $15,850,000 | $21,750,000 | 93% |

### Forecast Summary:
- **Total Forecasted (3 months):** $54,150,000
- **Expected Growth:** +8.2% per month (continuing trend)
- **Confidence:** High (based on consistent historical pattern)

### Methodology:
- **Data Points:** 10 historical periods
- **Trend:** Linear growth (+$580k/month avg)
- **Seasonal Factor:** 1.15x (Q4 typically higher)
- **Model Accuracy:** R² = 0.92 (excellent fit)

### Assumptions:
- Historical growth pattern continues
- No major market disruptions
- Seasonal patterns remain consistent

**Math Tool Verified:** ✅
**Note:** Forecasts are estimates based on historical trends

Forecast with Different Periods: ​

Short-term (1-3 periods):

"forecast next month"
"predict next quarter"
  • Higher accuracy
  • Tighter confidence intervals

Medium-term (4-6 periods):

"forecast next 6 months"
"predict next 2 quarters"
  • Good accuracy
  • Moderate confidence intervals

Long-term (7-12 periods):

"forecast next year"
"predict next 4 quarters"
  • Lower accuracy
  • Wider confidence intervals

13. Specialized Domain Use Cases ​

The tool automatically detects domain-specific patterns and provides specialized analysis:

TRIFR (Total Recordable Injury Frequency Rate) Analysis ​

What it does: Calculates workplace safety metrics automatically when safety data is detected

Auto-Detection: System recognizes safety datasets by column patterns:

  • Injury/incident columns
  • Exposure hours columns
  • Safety rate columns
  • Recordable injury data

Query Examples:

"TRIFR by quarter"
"total recordable injury frequency rate"
"safety rate trend quarterly"
"injury frequency by business unit"
"TRIFR breakdown by quarter"

Sample Response:

markdown
## TRIFR Analysis: Quarterly Breakdown

**Calculation:** (Total Recordable Injuries × 200,000) / Total Exposure Hours
**Analysis Period:** 2024 (Q1-Q4)

| Quarter | Recordable Injuries | Exposure Hours | TRIFR | Status |
|---------|-------------------|----------------|-------|--------|
| Q1 2024 | 3 | 520,000 | 1.15 | ⚠️ Above target |
| Q2 2024 | 2 | 535,000 | 0.75 | ✅ Meeting target |
| Q3 2024 | 1 | 542,000 | 0.37 | ✅ Exceeding target |
| Q4 2024 | 2 | 528,000 | 0.76 | ✅ Meeting target |

**Annual TRIFR:** 0.76 (8 injuries / 2,125,000 hours)
**Target TRIFR:** 1.00
**Status:** ✅ Below target (24% better than goal)

**Trend:** Improving ↓ (-68% from Q1 to Q3)

### Industry Comparison:
- **Your TRIFR:** 0.76
- **Industry Average:** 1.20
- **Performance:** 37% better than industry

**Math Tool Verified:** ✅
**Calculation Audit Trail:** Available in logs

TRIFR by Business Unit:

markdown
## TRIFR by Business Unit - Q3 2024

| Business Unit | Injuries | Hours | TRIFR | vs Target |
|---------------|----------|-------|-------|-----------|
| Manufacturing | 3 | 350,000 | 1.71 | ⚠️ 71% above |
| Logistics | 1 | 125,000 | 1.60 | ⚠️ 60% above |
| Warehouse | 2 | 180,000 | 2.22 | 🔴 122% above |
| Office | 0 | 85,000 | 0.00 | ✅ Excellent |

**Action Required:**
- **Warehouse:** Immediate safety review needed
- **Manufacturing:** Monitor closely
- **Office:** Maintain current practices

**Math Tool Verified:** ✅

Win Rate / Conversion Rate Analysis ​

Auto-Detection: Recognizes status-based workflows (sales, recruiting, support)

Query Examples:

"win rate by salesperson"
"conversion rate by source"
"success rate by team"
"close rate analysis"

Churn Rate / Turnover Analysis ​

Auto-Detection: Recognizes employee or customer lifecycle data

Query Examples:

"employee turnover by department"
"churn rate by quarter"
"retention rate analysis"

Capacity Utilization ​

Auto-Detection: Recognizes resource usage data

Query Examples:

"utilization rate by resource"
"capacity analysis by month"
"occupancy rate trend"

14. Identifier Analysis ​

What it does: Detailed breakdown for specific person/entity

Examples:

"analyze jordan faas"
"breakdown for john smith"
"detailed view of sarah miller"
"show all data for product ABC"

Sample Response:

markdown
## Identifier Analysis: Jordan Faas

**Total Records:** 87 deals
**Total Value:** $10,525,000
**Average Deal Size:** $121,034

### Breakdown by Status:
| Status | Count | Total Amount | Percentage |
|--------|-------|--------------|------------|
| Closed Won | 24 | $3,450,000 | 32.8% |
| Qualified | 31 | $3,875,000 | 37.5% |
| Lead | 32 | $3,200,000 | 29.7% |

### Breakdown by Product Category:
| Category | Count | Total Amount |
|----------|-------|--------------|
| Enterprise Software | 35 | $5,250,000 |
| Cloud Services | 28 | $3,150,000 |
| Professional Services | 24 | $2,125,000 |

### Performance Metrics:
- **Win Rate:** 32.8% (24 closed won / 73 closed total)
- **Average Days to Close:** 45 days
- **Largest Deal:** $450,000 (ABC Corp - Enterprise Software)
- **Most Recent Close:** $125,000 (Oct 15, 2024)

**Math Tool Verified:** ✅

13. Ranking Analysis ​

What it does: Ranks entities by performance metrics

Examples:

"who has the most deals?"
"rank salespeople by revenue"
"top performers by win rate"
"leaderboard by total closed won"
"highest win rate by salesperson"

Sample Response:

markdown
## Leaderboard: Sales Performance

**Ranked by:** Total Closed Won Revenue
**Period:** 2024 YTD

| Rank | Sales Person | Closed Won | Total Revenue | Win Rate | Avg Deal |
|------|--------------|------------|---------------|----------|----------|
| 1 | Sarah Miller | 42 | $7,850,000 | 38.5% | $186,905 |
| 2 | Jordan Faas | 38 | $6,450,000 | 35.2% | $169,737 |
| 3 | John Smith | 35 | $5,250,000 | 32.1% | $150,000 |
| 4 | Alice Johnson | 31 | $4,875,000 | 30.4% | $157,258 |
| 5 | Bob Williams | 28 | $4,200,000 | 29.2% | $150,000 |
...

**Total Team Revenue:** $52,450,000
**Average Win Rate:** 32.8%
**Math Tool Verified:** ✅

14. Pivot Tables / Cross-Tabulation ​

What it does: Multi-dimensional analysis (rows × columns)

Examples:

"show sales by region and product"
"pivot amount by salesperson and status"
"cross-tab deals by quarter and category"
"breakdown by team and priority"

Sample Response:

markdown
## Pivot Table: Deal Amount by Sales Person × Status

| Sales Person | Closed Won | Qualified | Lead | Total |
|--------------|------------|-----------|------|-------|
| Sarah Miller | $7,850,000 | $5,200,000 | $3,150,000 | $16,200,000 |
| Jordan Faas | $6,450,000 | $3,875,000 | $3,200,000 | $13,525,000 |
| John Smith | $5,250,000 | $4,100,000 | $2,850,000 | $12,200,000 |
| Alice Johnson | $4,875,000 | $3,650,000 | $2,950,000 | $11,475,000 |
...
| **Grand Total** | **$52,450,000** | **$35,000,000** | **$48,500,000** | **$135,950,000** |

**Math Tool Verified:** ✅ All row/column totals verified

15. Conditional Counting ​

What it does: Count records matching specific conditions

Examples:

"count where status = closed won"
"how many deals over $100,000?"
"count active employees in engineering"
"number of tickets with high priority"

Sample Response:

markdown
## Conditional Count Analysis

**Condition:** Status = "Closed Won" AND Deal Amount > $100,000

**Result:** 287 records match

### Breakdown:
- **Total Records:** 1,247
- **Matches Condition:** 287 (23.0%)
- **Does Not Match:** 960 (77.0%)

### Additional Insights:
- **Average Deal Size (matched):** $215,432
- **Total Value (matched):** $61,829,000
- **Percentage of Total Revenue:** 45.5%

**Math Tool Verified:** ✅

Query Examples by Use Case ​

General Dataset Operations (ANY Data) ​

Basic Exploration:

"list all datasets"
"show columns"
"describe dataset schema"
"show me 10 records"
"what data is available?"

Grouping & Counting:

"group by [any column]"
"count by [category]"
"breakdown by [field]"
"total by [column]"

Sorting & Ranking:

"sort by [column] descending"
"show top 10 by [metric]"
"rank by [value]"
"highest [column] values"

Filtering:

"show records where [column] = [value]"
"exclude [category]"
"filter by [field]"
"show only [condition]"

Time Analysis:

"monthly trend"
"quarterly breakdown"
"yearly comparison"
"temporal analysis"

Sales & CRM Data ​

Pipeline Analysis:

"show all deals"
"group by status"
"breakdown by stage"
"pipeline value by salesperson"
"deals by lead source"

Performance Metrics:

"win rate by salesperson"
"conversion rate by source"
"top performers by revenue"
"average deal size by product"
"close rate by quarter"

Temporal Analysis:

"monthly revenue trend"
"quarterly sales growth"
"forecast next quarter"
"deals closing this month"
"year over year growth"

Calculated Metrics:

"calculate commission = amount * 0.15"
"win rate = closed won / total closed"
"average days to close"

HR & Employee Data ​

Headcount Analysis:

"count employees by department"
"headcount by location"
"group by level"
"breakdown by manager"
"employee distribution"

Compensation Analysis:

"average salary by level"
"compensation by department"
"salary distribution"
"group salary by role"

Turnover & Retention:

"turnover rate by department"
"employee tenure distribution"
"new hires by quarter"
"employees with less than 1 year"
"retention rate analysis"

Temporal Trends:

"hiring trend by month"
"quarterly headcount growth"
"turnover trend yearly"
"compensation growth rate"

Inventory & Supply Chain ​

Stock Analysis:

"products below reorder point"
"stock levels by warehouse"
"items with zero quantity"
"low stock alerts"
"out of stock items"

Value Analysis:

"total inventory value by category"
"average price by supplier"
"most expensive items"
"inventory value trend"

Movement & Turnover:

"turnover rate by product"
"products expiring this month"
"fast-moving items"
"slow-moving inventory"

Temporal Patterns:

"weekly stock levels"
"monthly inventory value"
"quarterly turnover rate"
"seasonal demand patterns"

Customer Support & Tickets ​

Ticket Analysis:

"tickets by priority"
"count by status"
"group by support agent"
"breakdown by category"
"escalated tickets"

Performance Metrics:

"average resolution time by team"
"response time by priority"
"customer satisfaction by agent"
"tickets per agent"

Status & Workflow:

"open vs closed tickets"
"unresolved tickets over 3 days"
"tickets in progress"
"completed this week"

Temporal Trends:

"ticket volume by week"
"monthly resolution time trend"
"quarterly escalation rate"
"forecast next month volume"

Safety & TRIFR Data ​

Safety Metrics:

"TRIFR by quarter"
"injury rate by business unit"
"recordable incidents quarterly"
"safety rate trend"
"total recordable injury frequency rate"

Temporal Analysis:

"quarterly safety breakdown"
"monthly incident trend"
"yearly TRIFR comparison"
"safety improvement over time"

Business Unit Comparison:

"TRIFR by department"
"injury rate by location"
"safety performance by site"
"highest risk units"

Exposure Hours:

"total exposure hours by quarter"
"hours worked by business unit"
"exposure trend monthly"

Financial & Accounting Data ​

Revenue Analysis:

"revenue by product line"
"sales by region"
"income by category"
"top revenue sources"

Expense Analysis:

"expenses by department"
"cost breakdown by category"
"top expenses this year"
"spending by vendor"

Profitability:

"profit margin by quarter"
"gross profit by product"
"calculate profit = revenue - cost"
"margin trend analysis"

Budget & Variance:

"budget vs actual by month"
"variance analysis by department"
"spending over budget"
"budget utilization rate"

Temporal Financial Trends:

"quarterly revenue growth"
"monthly expense trend"
"cash flow by quarter"
"forecast next quarter revenue"
"year over year comparison"

Manufacturing & Production ​

Production Metrics:

"units produced by line"
"production by shift"
"output by machine"
"daily production trend"

Quality Metrics:

"defect rate by product"
"quality score by line"
"rework rate by department"
"first-pass yield"

Efficiency:

"cycle time by product"
"downtime by machine"
"utilization rate by line"
"efficiency trend monthly"

Temporal Production:

"weekly production volume"
"monthly output trend"
"quarterly quality metrics"
"forecast production capacity"

Academic & Education Data ​

Student Performance:

"average grade by department"
"group by major"
"student count by program"
"GPA distribution"

Course Analysis:

"enrollment by course"
"completion rate by department"
"average score by instructor"
"course popularity trend"

Temporal Education:

"enrollment trend by semester"
"quarterly graduation rate"
"yearly performance comparison"
"student retention by year"

Best Practices ​

1. Start Simple, Then Refine ​

❌ Don't start with:

"show me the total closed won deals by salesperson excluding closed lost for Q3 2024 with commission calculations"

✅ Instead, build up:

"group by salesperson | status=closed won"
"show results for Q3 2024"
"calculate commission = deal amount * 0.15"

2. Use Filters with Context ​

Format: query | filter1=value1, filter2=value2

Examples:

"sum amount | status=closed won"
"group by salesperson | status=closed won, date=2024-01-01 to 2024-12-31"
"show records | exclude status=closed lost, category=inactive"

3. Leverage Fuzzy Matching ​

You don't need exact spellings:

"group by jordan" instead of "group by Jordan Faas"
"sum by tech" instead of "sum by Tech Support"
"show engineering" instead of "show department=Engineering"

4. Ask for Schema First ​

When exploring new datasets:

"list columns"                    # See what's available
"show dataset schema"             # Get detailed view
"summarize deal amounts"          # Understand data distribution
"group by status"                 # See categories

5. Combine Operations ​

You can chain multiple operations:

"show top 20 sorted by amount with commission = amount * 0.15"
"group by status | exclude closed lost | sort by count descending"
"sum amount by salesperson | status=closed won | date=2024"

6. Use Explicit Exclusions ​

❌ Unclear:

"show active deals"

✅ Explicit:

"show records | exclude status=closed lost, exclude status=disqualified"

7. Verify Calculations ​

All calculations are automatically verified with MathTool, but you can request explicit verification:

"sum amount by status and verify totals"
"calculate commission and show verification"
"group by salesperson with math verification"

Troubleshooting ​

Problem: "No datasets available" ​

Cause: No data loaded in session

Solution:

  1. Verify you have datasets uploaded
  2. Check that chat_with_content includes dataset IDs
  3. List datasets: "show all datasets"

Problem: "Column not found" ​

Cause: Column name doesn't match dataset

Solution:

  1. Check available columns: "list columns"
  2. Try fuzzy matching: "amount" instead of "Deal Amount"
  3. Use schema: "show dataset schema"

Problem: "Multiple matches found for value" ​

Cause: Fuzzy value matching found ambiguous results

Example:

Query: "group by smith"
Error: Found 3 matches: "John Smith", "Jane Smith", "Bob Smith"

Solution: Be more specific:

"group by john smith"
"group by jane smith"

Problem: "No match found for value" ​

Cause: Value doesn't exist in any column

Solution:

  1. Check spelling
  2. View sample values: "show dataset schema"
  3. List unique values: "group by [column]"

Problem: Query times out ​

Cause: Query too complex or dataset too large

Solution:

  1. Add filters to reduce dataset size
  2. Break into smaller queries
  3. Use specific column names instead of fuzzy matching
  4. Limit results: "show top 100"

Problem: Unexpected grouping column ​

Cause: Fuzzy matching selected wrong column

Solution: Be explicit:

"group by status column"
"sum amount column by salesperson column"

Problem: Calculation seems wrong ​

Cause: Possible data quality issues

Solution:

  1. All calculations are MathTool-verified (50-digit precision)
  2. Check for null values: "show dataset schema"
  3. Verify filters applied correctly
  4. Request explicit verification: "verify calculation"

Domain-Agnostic Design Principles ​

How It Works Automatically ​

The tool is truly domain-agnostic - it works with ANY dataset without hardcoded assumptions:

1. Auto-Detection of Column Types ​

  • Date columns: Any column with dates, timestamps, or "date" in the name
  • Numeric columns: Amount, value, price, quantity, count, hours, etc.
  • Entity columns: People, products, departments, categories, etc.
  • Status columns: Active/inactive, open/closed, success/failure, etc.

2. Intelligent Column Selection ​

When you don't specify columns, the system automatically:

  • Finds the best date column for temporal analysis
  • Selects the most meaningful numeric column for calculations
  • Identifies the optimal grouping column based on cardinality (2-50 unique values)
  • Avoids ID columns and high-cardinality fields

3. Context-Aware Analysis ​

The tool adapts its analysis based on your data:

  • Sales Data: Focuses on revenue, deals, win rates
  • HR Data: Analyzes headcount, turnover, compensation
  • Safety Data: Calculates TRIFR, injury rates, exposure hours
  • Inventory: Tracks stock levels, turnover, values
  • ANY Custom Data: Discovers patterns automatically

4. No Configuration Required ​

  • Works with any column names (not just "Amount" or "Status")
  • Handles any date formats automatically
  • Adapts to any business domain
  • No setup, no schema definitions, no mappings

Example: Same Query, Different Datasets ​

The query "show monthly trend" works across all datasets:

Sales Dataset:

  • Groups by: Close Date
  • Calculates: Total Deal Amount, Deal Count
  • Shows: Revenue trend over time

HR Dataset:

  • Groups by: Hire Date
  • Calculates: Headcount, New Hires
  • Shows: Hiring trend over time

Safety Dataset:

  • Groups by: Incident Date
  • Calculates: TRIFR, Injury Count, Exposure Hours
  • Shows: Safety metrics over time

Same query. Different insights. Zero configuration.

Summary ​

The Data Analytics tool transforms natural language questions into precise data analysis across any dataset. With features ranging from basic grouping to advanced forecasting, temporal trending, TRIFR calculations, fuzzy matching, and financial-grade accuracy, it provides powerful insights without requiring SQL knowledge or manual configuration.

Key Strengths ​

✅ Natural Language - Ask questions in plain English
✅ Truly Domain Agnostic - Works with ANY dataset automatically (sales, HR, safety, inventory, custom)
✅ Fuzzy Matching - Understands partial names and values ("jordan" → "Jordan Faas")
✅ Financial Precision - 50-digit accuracy with MathTool verification
✅ Temporal Analysis - Trends, forecasts, growth rates (hourly to yearly)
✅ Specialized Metrics - Automatic TRIFR, win rates, turnover calculations
✅ Smart Auto-Detection - Automatically finds date, value, and entity columns
✅ Rich Analysis - 15+ analysis types including pivots, percentages, rankings
✅ Clear Results - Formatted markdown tables with audit trails

Complete Feature List ​

Basic Operations:

  1. Dataset Discovery
  2. Column Inspection
  3. Grouping & Aggregation
  4. Sorting & Ranking
  5. Filtering (basic, comparison, exclusion)

Analysis Features:

  1. Statistical Analysis (min, max, mean, median, std dev)
  2. Percentages & Ratios
  3. Calculated Columns
  4. Fuzzy Value Matching
  5. Temporal Analysis (hourly, daily, weekly, monthly, quarterly, yearly)
  6. Trending Analysis (growth rates, momentum, patterns)
  7. Forecasting (1-12 periods ahead with confidence intervals)
  8. Specialized Domain Metrics (TRIFR, win rates, churn, etc.)
  9. Identifier Analysis (detailed entity breakdowns)
  10. Ranking Analysis (leaderboards, performance metrics)
  11. Pivot Tables (multi-dimensional analysis)
  12. Conditional Counting

What's Next? ​

Explore your data with natural language queries and discover insights instantly. The tool learns from your data structure and adapts automatically - no configuration needed!

Quick Start:

"list datasets"              # See what's available
"list columns"               # Understand the structure
"show monthly trend"         # See temporal patterns
"group by [column]"          # Get category insights
"forecast next 3 months"     # Predict future trends
"show top 10 by [metric]"    # Find top performers

Advanced Queries:

"TRIFR by quarter | exclude business unit=office"
"win rate by salesperson | status=closed"
"monthly revenue trend | exclude status=cancelled"
"forecast next 6 months quarterly"
"group by department with average salary"

Questions? Issues? All operations are logged to CloudWatch for debugging and audit trails with full MathTool verification.