XDF Stream Selector

A Python toolkit for extracting, filtering, synchronizing, and resampling multi-modal data streams from XDF (Extensible Data Format) files, plus utilities for survey data cleanup.

Features

XDF Processing

• 🔍 Robust name matching - Case-insensitive and whitespace-tolerant stream/channel selection
• 🔄 Multi-stream synchronization - Align streams with different sampling rates
• ⚡ Flexible resampling - Interpolate data to any target frequency
• ✂️ Edge truncation - Remove unstable data from recording boundaries
• 🛠️ Diagnostic tools - Debug stream selection issues easily
• 📊 All-in-one pipeline - Process everything with a single function call

Survey Data Processing

• 🧹 Automatic cleanup - Remove invalid entries and format inconsistencies
• 🔢 Smart type conversion - Handle numeric fields with commas and special characters
• 📅 Year-to-experience conversion - Automatically convert birth years to experience values
• ✅ Data validation - Filter out non-numeric participant IDs

Installation

pip install pandas numpy scipy pyxdf

Then download xdf_stream_processing.py and place it in your project directory.

Quick Start

XDF Processing

importpyxdffromxdf_stream_processingimportprocess_xdf_streams, print_stream_info# Load your XDF filestreams, header=pyxdf.load_xdf("recording.xdf")
# Inspect available streamsprint_stream_info(streams)
# Define what you want to extractselection= {
"VarjoEyeMetrics": ["leftPupilDiam", "rightPupilDiam"],
"Polar H10": ["HR", "RRI"],
}
# Process everything in one calldf=process_xdf_streams(
streams=streams,
selection_dict=selection,
target_freq=50.0, # Resample to 50 Hztruncate_n=100, # Remove 100 samples from each endverbose=True
)
# Save your processed datadf.to_csv("processed_data.csv", index=False)

Usage

1. Inspect Your Data

fromxdf_stream_processingimportprint_stream_info, get_copyable_format# See all available streams and channelsprint_stream_info(streams)
# Get a copyable template for selection_dictget_copyable_format(streams)

2. Select and Process Streams

The main function handles everything:

df=process_xdf_streams(
streams=streams, # From pyxdf.load_xdf()selection_dict=selection, # What to extracttarget_freq=50.0, # Target sampling rate (Hz)truncate_n=100, # Edge samples to removeverbose=True# Show progress
)

3. Custom Workflows

For more control, use individual functions:

fromxdf_stream_processingimport (
select_streams_by_name,
streams_to_dataframe,
resample_dataframe,
truncate_dataframe
)
# Step-by-step processingselected=select_streams_by_name(streams, selection)
df=streams_to_dataframe(selected, target_freq=50.0)
df=resample_dataframe(df, target_freq=100.0)
df=truncate_dataframe(df, n=50)

Survey Data Processing

The toolkit includes utilities for cleaning survey data, particularly useful for preprocessing questionnaire responses before analysis.

Features

• Participant ID cleanup: Remove special characters (e.g., '#5' → '5')
• Invalid entry filtering: Remove non-numeric participant IDs (e.g., rows starting with letters)
• Numeric field cleaning: Remove commas from numeric fields (e.g., '1,000' → '1000')
• Smart year conversion: Convert year values to experience (e.g., '1995' → '30' for experience)

Usage Example

importpandasaspd# Load your survey datasurvey_data=pd.read_csv("raw_survey.csv")
# Make a proper copy to avoid warningsselected_data=survey_data.copy()
# Clean participant IDs - remove '#' prefixselected_data['pid'] =selected_data['pid'].astype(str).str.replace('#', '', regex=False)
# Filter out invalid participant IDs (keep only numeric)selected_data=selected_data[selected_data['pid'].str.match(r'^\d', na=False)].copy()
# Remove commas from numeric fields (e.g., flight hours)selected_data['flt_hrs'] =selected_data['flt_hrs'].astype(str).str.replace(',', '', regex=False)
# Clean experience column with smart year detectiondefclean_experience(val):
try:
num=float(val.replace(',', ''))
ifnum>=1900: # Likely a year, convert to experiencereturnstr(int(2025-num))
else: # Already an experience valuereturnstr(int(num))
except:
returnval# Keep original if conversion failsselected_data['experience'] =selected_data['experience'].astype(str).apply(clean_experience)
# Save cleaned dataclean_survey_data=selected_data.copy()
clean_survey_data.to_csv('clean_survey_data.csv', index=False)
print(f"Cleaned {len(clean_survey_data)} survey responses")

Common Survey Cleanup Steps

1. Remove special characters from IDs

   df['pid'] =df['pid'].astype(str).str.replace('#', '', regex=False)

2. Filter invalid entries

   df=df[df['pid'].str.match(r'^\d', na=False)].copy()

3. Clean numeric fields

   df['numeric_field'] =df['numeric_field'].astype(str).str.replace(',', '', regex=False)

4. Convert years to age/experience

   defyear_to_experience(val, current_year=2025):
   try:
   num=float(val.replace(',', ''))
   ifnum>=1900:
   returnstr(int(current_year-num))
   returnstr(int(num))
   except:
   returnvaldf['experience'] =df['experience'].astype(str).apply(year_to_experience)

Validation Checks

After cleaning, always verify your data:

print(f"Shape after cleaning: {clean_survey_data.shape}")
print(f"\nUnique participant IDs: {clean_survey_data['pid'].nunique()}")
print(f"\nCleaned experience values:\n{clean_survey_data['experience'].value_counts()}")

Key Parameters

XDF Processing

Parameter	Description	Default
target_freq	Target sampling frequency in Hz	50.0
truncate_n	Samples to remove from each end	0
verbose	Print detailed progress information	True
use_timestamps	Use original timestamps vs synthetic	True

Selection Dictionary Format

The selection_dict maps stream names to lists of channel names:

selection_dict= {
"StreamName1": ["channel1", "channel2", "channel3"],
"StreamName2": ["channelA", "channelB"],
# Add more streams as needed
}

Features:

• ✅ Case-insensitive matching
• ✅ Whitespace-tolerant
• ✅ Partial matches allowed (extracts available channels)
• ✅ Special handling for Polar H10 devices

Output Format

The processed DataFrame contains:

• Time column: Common time base in seconds
• {StreamName}_{ChannelName} columns: One per selected channel

Example:

 Time VarjoEyeMetrics_leftPupilDiam VarjoEyeMetrics_rightPupilDiam Polar H10_HR
0 0.000 3.245 3.156 72.3
1 0.020 3.248 3.159 72.3
2 0.040 3.251 3.162 72.4
...

Common Use Cases

XDF Streams

High-frequency signals (ECG, EEG)

df=process_xdf_streams(streams, selection, target_freq=250.0)

Low-frequency signals (HR, pupil diameter)

df=process_xdf_streams(streams, selection, target_freq=10.0)

No resampling (keep original rates)

df=streams_to_dataframe(streams, resample=False)

Remove edge effects

df=process_xdf_streams(streams, selection, truncate_n=200)

Survey Data

Clean pilot study data

# Remove test entries and format fieldssurvey=survey[survey['pid'].str.match(r'^\d', na=False)].copy()
survey['flt_hrs'] =survey['flt_hrs'].str.replace(',', '')

Prepare for statistical analysis

# Convert all numeric fields to proper typesnumeric_cols= ['flt_hrs', 'experience', 'age']
forcolinnumeric_cols:
survey[col] =pd.to_numeric(survey[col], errors='coerce')

Examples

See example_usage.py for comprehensive examples including:

1. Basic workflow with synthetic data
2. Loading and processing real XDF files
3. Custom processing workflows
4. Diagnostic and debugging tools
5. Multiple processing frequencies
6. Survey data cleanup workflows

Run examples:

python example_usage.py

Functions Overview

XDF Processing

Function	Purpose
process_xdf_streams()	Complete pipeline (recommended)
print_stream_info()	Display available streams/channels
get_copyable_format()	Generate selection template
select_streams_by_name()	Filter streams and channels
streams_to_dataframe()	Synchronize and merge streams
resample_dataframe()	Resample to uniform frequency
truncate_dataframe()	Remove edge samples

Survey Processing

Survey data cleanup functions are provided as code snippets in the documentation above. For a complete implementation, see the examples directory.

Requirements

• Python 3.7+
• pandas
• numpy
• scipy
• pyxdf (for loading XDF files)

Citation

If you use XDF Stream Selector in your research, please cite:

@software{xdf_stream_selector,
author = {Md Mijanur Rahman, Niklas P. Schulmeyer},
title = {XDF Stream Selector: A Python Toolkit for Multi-Modal Data Stream Processing},
year = {2025},
url = {https://github.com/aeon-lab/stream_selector},
version = {1.0.0}
}

License

This project is licensed under the MIT License - see below for details.

MIT License
Copyright (c) 2025 [Your Name/Organization]
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

Authors

Md Mijanur Rahman
Niklas P. Schulmeyer

Contributing

Contributions welcome! Please feel free to submit a Pull Request.