Export and Save: Working with Generated Time Series

Phoenix offers multiple ways to persist and export your generated time series data, enabling integration with SENTINEL, external analysis tools, and long-term storage.

Two Workflows: Save vs. Download

Phoenix provides two distinct workflows for working with generated data:

Save to Database

Permanent storage in Kronts database
Tracked under your user account
Accessible in SENTINEL, FORGE, CEREBRO
Regenerable with same parameters
Subject to limits: 3 series per user, 10,000 points max

Download to File

Immediate export to local file
No storage in database
Portable for external tools
No user limits on number of downloads
Three formats: CSV, Excel, JSON

Saving Time Series to Database

When to Save

Save When: - You'll use data in SENTINEL for analysis - You want to preserve generation parameters - You'll reference this data in FORGE/CEREBRO - You need team access to the dataset - You want to regenerate with slight variations later

Don't Save When: - Quick one-time testing - Already have 3 saved series (limit reached) - Just need data in external tool (use download instead) - Exceeds 10,000 point limit

Save Workflow

Step 1: Generate and Preview

Before saving, always preview to verify the time series looks correct:

Configure parameters
Click "Preview"
Verify chart and statistics
Check for warnings (aliasing, etc.)

Important: Never save without previewing first!

Step 2: Click "Save Time Series"

Locate the "Save Time Series" button: - Top of sidebar (above configuration) - Bottom of sidebar (after degradation options) - OR scroll to chart area (below chart)

Click the button to open the save modal.

[Screenshot Required: Save Button Location] 1. After previewing a time series 2. Capture: Full interface showing save button locations 3. Annotate: Circle the save buttons 4. Purpose: Help users find save functionality

The save modal appears with three fields:

Time Series Name (Required) - Descriptive name for this dataset - Will appear in lists and dashboards - Best practice: Include key parameters

Good Names:

"Motor_1800RPM_30Hz_Clean"
"Temp_DailyCycle_5deg_2weeks"
"3Axis_Vibration_Correlated_0.85"
"Pressure_Trend_With_10pct_Outliers"

Poor Names:

"Test"
"Data1"
"My Time Series"
"Series_20240115" (without context)

Description (Optional but Recommended) - Detailed notes about the time series - Generation parameters - Intended use case - Degradation details if applicable

Good Description:

"3-channel temperature sensor array, high correlation (0.88-0.92),
daily oscillation (5°C amplitude), 2% missing data, 1% outliers at 9999.
For testing sensor fusion algorithm in FORGE."

Open in Sentinel (Checkbox, Optional) - If checked: Immediately redirects to SENTINEL with this time series - If unchecked: Returns to Phoenix generation page

[Screenshot Required: Save Modal] 1. Click "Save Time Series" 2. Capture: Modal dialog with all fields visible 3. Purpose: Show save interface and fields

Step 4: Save and Confirm

Click "Save" button in modal.

Phoenix will: 1. Validate user limits (max 3 series) 2. Regenerate time series (ensures consistency) 3. Validate point count (max 10,000) 4. Store in database with metadata 5. Show success message 6. Redirect (if "Open in Sentinel" checked)

Success: Confirmation message appears Failure: Error message explains issue (see Troubleshooting)

What Gets Saved

Data Stored: - All data points (timestamps + values) - Channel names and units (multi-channel) - Time series name and description - Generation parameters (full configuration) - Creation timestamp - User association

Generation Parameters Include: - Time configuration (duration, sampling frequency) - Base signal (mean, noise, trend) - All oscillations (frequency, amplitude, phase) - Channel configurations (multi-channel) - Correlations (multi-channel) - Degradation settings (removal, outliers)

Why Parameters Matter: - Reproduce same signal exactly - Generate variations (tweak one parameter) - Document methodology - Audit trail

Accessing Saved Time Series

In Phoenix: - Navigate to /phoenix/ (home page) - Your saved series listed - Click series to regenerate with same parameters - Edit parameters and preview modifications

In SENTINEL: - Select saved time series from dashboard - Analyze data quality automatically - View insights and recommendations

In FORGE/CEREBRO: - Import from saved Phoenix series - Process or label as needed

Regenerating from Saved Series

Why Regenerate?

Use Cases: - Create variations (change one parameter) - Generate longer/shorter versions - Add/remove degradation - Test parameter sensitivity - Create family of related datasets

Regeneration Workflow

Navigate to /phoenix/ (home page)
Find saved time series in list
Click time series name or "Regenerate" button
Phoenix loads generation form with all original parameters
Make desired changes
Preview new version
Save as new series (or download)

Example: Creating Parameter Sweep

Original Series: "Motor_1800RPM_Clean" - Motor at 1800 RPM (30 Hz)

Variations: 1. Regenerate → Change to 1500 RPM (25 Hz) → Save as "Motor_1500RPM_Clean" 2. Regenerate → Change to 2100 RPM (35 Hz) → Save as "Motor_2100RPM_Clean" 3. Regenerate → Add 5% outliers → Save as "Motor_1800RPM_5pct_Outliers"

Now you have a family of related test datasets!

Managing Saved Series

Viewing: - Home page lists all your saved series - Shows name, creation date, point count

Deleting: - Click delete button (⚠️ No undo!) - Frees up limit slot (for users at 3-series limit) - Removes from database permanently

Editing Metadata: - Currently not supported - Workaround: Regenerate and save with new name/description - Delete old version if needed

Downloading Time Series Files

When to Download

Download When: - Analyzing in Python, R, MATLAB, Excel - Sharing with external collaborators - Archiving for long-term storage - Already at save limit (3 series) - No need for Kronts integration

Download Formats

Phoenix offers three export formats:

CSV (Comma-Separated Values)

Best For: - Python, R, MATLAB analysis - Excel/Google Sheets import - Universal compatibility - Simple text format

Structure (Single-Channel):

Timestamp,value
2024-01-15 10:00:00,100.5
2024-01-15 10:00:01,101.2
2024-01-15 10:00:02,99.8

Structure (Multi-Channel):

Timestamp,X-Axis,Y-Axis,Z-Axis
2024-01-15 10:00:00,0.5,0.3,-9.8
2024-01-15 10:00:01,0.6,0.4,-9.7
2024-01-15 10:00:02,0.4,0.2,-9.9

File Naming: time_series_YYYY-MM-DD_HH-MM-SS.csv

Excel (XLSX)

Best For: - Direct Excel analysis - Non-technical users - Formatted spreadsheets - Charts in Excel

Structure: Same data as CSV, but in Excel format

Advantages: - Preserves data types - Can include multiple sheets (future enhancement) - Better for Windows environments

File Naming: time_series_YYYY-MM-DD_HH-MM-SS.xlsx

JSON

Best For: - API integration - Structured data with metadata - Programming language compatibility - Includes generation parameters

Structure:

{
  "metadata": {
    "name": "Generated Time Series",
    "created_at": "2024-01-15T10:00:00Z",
    "sampling_frequency": 1.0,
    "channels": ["value"]
  },
  "statistics": {
    "value": {
      "min": 85.2,
      "max": 114.8,
      "mean": 99.9,
      "count": 3600
    }
  },
  "data": [
    {"timestamp": "2024-01-15T10:00:00", "value": 100.5},
    {"timestamp": "2024-01-15T10:00:01", "value": 101.2},
    ...
  ]
}

Advantages: - Includes metadata and statistics - Self-documenting - Easy to parse in any language - Structured format

File Naming: time_series_YYYY-MM-DD_HH-MM-SS.json

Download Workflow

Step 1: Generate and Preview

Always preview before downloading: 1. Configure parameters 2. Click "Preview" 3. Verify chart looks correct

Locate the "Download" button: - Near the "Save" button - Shows dropdown arrow (▼)

Click to reveal format options.

[Screenshot Required: Download Dropdown] 1. After previewing a time series 2. Click download button to show dropdown 3. Capture: Dropdown menu showing CSV, XLSX, JSON options 4. Purpose: Show download options

Step 3: Select Format

Click desired format: - CSV - Most common, universal - Excel (XLSX) - For Excel users - JSON - For API/programming

File downloads immediately to your browser's download folder.

Step 4: Import to External Tool

Python (pandas):

import pandas as pd

# CSV
df = pd.read_csv('time_series_2024-01-15_10-00-00.csv', parse_dates=['Timestamp'])
df.set_index('Timestamp', inplace=True)

# Excel
df = pd.read_excel('time_series_2024-01-15_10-00-00.xlsx', parse_dates=['Timestamp'])
df.set_index('Timestamp', inplace=True)

# JSON
df = pd.read_json('time_series_2024-01-15_10-00-00.json', orient='records')

# CSV
df <- read.csv('time_series_2024-01-15_10-00-00.csv')
df$Timestamp <- as.POSIXct(df$Timestamp)

# Excel
library(readxl)
df <- read_excel('time_series_2024-01-15_10-00-00.xlsx')

MATLAB:

% CSV
df = readtable('time_series_2024-01-15_10-00-00.csv');

% Excel
df = readtable('time_series_2024-01-15_10-00-00.xlsx');

Excel: - Open file directly - Data → From Text/CSV (for CSV import) - File → Open (for XLSX)

Integration with SENTINEL

Direct Integration: "Open in Sentinel"

When saving a time series, check "Open in Sentinel" to: 1. Save time series to database 2. Immediately redirect to SENTINEL 3. Auto-load the time series for analysis

SENTINEL Features: - Automatic data quality dashboard - Statistical analysis - Data characteristics visualization - Quality recommendations

Use Case: - Generate test data with known characteristics - Validate SENTINEL's quality detection - Benchmark quality metrics - Train users on interpreting quality reports

Workflow: Phoenix → SENTINEL → FORGE

Complete Pipeline:

Phoenix: Generate clean data → Save
Phoenix: Generate degraded version → Save
SENTINEL: Analyze degraded data → Identify issues
FORGE: Clean data based on recommendations
Compare: Cleaned data vs. original clean data
Measure: Cleaning effectiveness

This validates your entire data quality workflow!

User Limits and Quotas

Save Limits

Regular Users: - Maximum 3 saved time series - Must delete old series to save new ones - No limit on downloads

Superusers/Admins: - Unlimited saved series - No quotas

Point Limits

All Users: - Maximum 10,000 total data points - Single-channel: 10,000 points per series - Multi-channel: 10,000 total across all channels

Example Calculations:

Single-channel, 1 Hz, 2.78 hours = 10,000 points ✓
3 channels, 1 Hz, 55 minutes = 9,900 points ✓
10 channels, 10 Hz, 100 seconds = 10,000 points ✓

Troubleshooting

"Maximum series limit reached"

Cause: You have 3 saved time series (regular user limit)

Solutions: 1. Delete an old saved series 2. Use download instead of save 3. Contact admin for increased limit

"Data point limit exceeded"

Cause: Trying to save > 10,000 total points

Solutions: 1. Reduce duration 2. Reduce sampling frequency 3. Reduce number of channels (multi-channel) 4. Calculate limit: Duration × Frequency × Channels ≤ 10,000

Example Fix:

Problem: 3 channels × 1 hour × 1 Hz = 10,800 points ✗
Fix: Reduce to 55 minutes: 3 × 3300 × 1 = 9,900 points ✓

Save succeeds but can't find saved series

Check: 1. Navigate to /phoenix/ home page (not /phoenix/generate/) 2. Verify you're logged in as same user 3. Check time series list on home page 4. Series sorted by creation date (newest first)

Download button doesn't work

Check: 1. Preview generated first (download requires preview) 2. Check browser popup blocker settings 3. Verify JavaScript enabled 4. Try different format (CSV vs XLSX vs JSON)

Downloaded file is empty or corrupted

Causes: 1. No data generated (preview first) 2. Browser cancelled download 3. Disk space full

Solutions: - Always preview before downloading - Check browser download status - Verify disk space - Try again

Can't import CSV into Python/R

Common Issues:

Timestamp parsing:

# Add parse_dates parameter
df = pd.read_csv('file.csv', parse_dates=['Timestamp'])

Column names:

# Check exact column names
print(df.columns)
# Use correct name (case-sensitive)

Encoding:

# Try specifying encoding
df = pd.read_csv('file.csv', encoding='utf-8')

JSON format doesn't parse

Check: 1. Valid JSON (use online validator) 2. Correct orientation/structure 3. Large files may need streaming parser

Best Practices

Naming Conventions

Include Key Information: - Signal type - Key parameters - Degradation if applicable - Use case

Format Suggestions:

{Type}_{Parameter}_{Quality}_{Purpose}

Examples:
Vibration_30Hz_Clean_Testing
Temp_DailyCycle_5pct_Outliers_FORGE
Pressure_150kPa_Drift_Validation

Documentation in Descriptions

Template:

Signal: {type}
Parameters: {key settings}
Channels: {single/multi-channel details}
Degradation: {removal%, outlier%}
Purpose: {intended use}
Notes: {additional context}

Example:

Signal: 3-axis accelerometer
Parameters: 30 Hz main oscillation, 2 m/s² amplitude
Channels: X, Y, Z correlated at 0.35
Degradation: 3% removal, 2% outliers (constant 999)
Purpose: Test vibration analysis algorithm robustness
Notes: Simulates motor at 1800 RPM with bearing fault

Version Control

Strategy: Use names with version numbers

Motor_1800RPM_v1_Clean
Motor_1800RPM_v2_Light_Degradation
Motor_1800RPM_v3_Heavy_Degradation

Export for Archival

Workflow: 1. Save to database for immediate use 2. Download JSON for long-term archive 3. JSON includes metadata and parameters 4. Store with documentation

Advanced Workflows

Batch Generation

While Phoenix doesn't support batch generation directly:

Manual Batch Workflow: 1. Generate first variant → Save/Download 2. Regenerate with modified parameter → Save/Download 3. Repeat for all variants 4. Organize saved series with clear naming

External Post-Processing

Workflow: 1. Generate in Phoenix 2. Download CSV 3. Load in Python/R 4. Apply additional processing: - Custom degradation patterns - Complex transformations - Merge with other data 5. Save processed version

Example (Python):

import pandas as pd

# Load Phoenix CSV
df = pd.read_csv('phoenix_output.csv', parse_dates=['Timestamp'])

# Add custom processing
df['processed'] = df['value'].rolling(window=10).mean()

# Add systematic gaps (every 100th point)
df_systematic = df[df.index % 100 != 0]

# Save for reimport
df_systematic.to_csv('processed_data.csv', index=False)

Multi-Format Archiving

Best Practice: Export all formats for flexibility

Workflow: 1. Preview time series 2. Download CSV (for analysis) 3. Download JSON (for metadata preservation) 4. Download XLSX (for stakeholder sharing) 5. Save to database (for Kronts integration)

Summary

Phoenix offers flexible export and save options:

Save to Database: - Permanent storage in Kronts - SENTINEL/FORGE/CEREBRO integration - Regeneration with parameters - User limits: 3 series, 10,000 points

Download Files: - CSV: Universal, simple - XLSX: Excel-friendly - JSON: Metadata-rich, structured - No limits on downloads

Best Practices: - Preview before save/download - Use descriptive names - Document in descriptions - Archive in multiple formats - Validate imports in external tools

Next Steps

Basic Usage - Review complete generation workflow
Multi-Channel - Export multi-channel data
Data Degradation - Add quality issues before export
Technical Reference - File format specifications

Proper export and save workflows ensure your synthetic time series data integrates seamlessly with analysis tools and the broader Kronts ecosystem.

Export and Save: Working with Generated Time Series

Two Workflows: Save vs. Download

Save to Database

Download to File

Saving Time Series to Database

When to Save

Save Workflow

Step 1: Generate and Preview

Step 2: Click "Save Time Series"

Step 3: Fill Save Modal

Step 4: Save and Confirm

What Gets Saved

Accessing Saved Time Series

Regenerating from Saved Series

Why Regenerate?

Regeneration Workflow

Managing Saved Series

Downloading Time Series Files

When to Download

Download Formats

CSV (Comma-Separated Values)

Excel (XLSX)

JSON

Download Workflow

Step 1: Generate and Preview

Step 2: Click Download Dropdown

Step 3: Select Format

Step 4: Import to External Tool

Integration with SENTINEL

Direct Integration: "Open in Sentinel"

Workflow: Phoenix → SENTINEL → FORGE

User Limits and Quotas

Save Limits

Point Limits

Troubleshooting

"Maximum series limit reached"

"Data point limit exceeded"

Save succeeds but can't find saved series

Download button doesn't work

Downloaded file is empty or corrupted

Can't import CSV into Python/R

JSON format doesn't parse

Best Practices

Naming Conventions

Documentation in Descriptions

Version Control

Export for Archival

Advanced Workflows

Batch Generation

External Post-Processing

Multi-Format Archiving

Summary

Next Steps