Python WebRTC Models

0.3.0 · active · verified Thu Apr 16

A Python library providing lightweight models designed for analyzing WebRTC statistics. It helps estimate quality of experience, predict network congestion, and perform other related tasks using simple prediction models. Currently at version 0.3.0, it follows a feature-driven release cadence, with updates for new models or API refinements.

Common errors

```
TypeError: PacketLossPredictor.predict() got an unexpected keyword argument 'inbound_rtp_packets_lost'
```
cause The `predict` method signature changed in version 0.3.0 from accepting keyword arguments to a single dictionary.

fix Pass a dictionary of features instead: `predictor.predict({'inbound_rtp_packets_lost': 0, 'inbound_rtp_packets_received': 100})`.
```
KeyError: 'some_missing_feature'
```
cause The input dictionary to `predict` is missing a required feature key or has a misspelled key for the specific model being used.

fix Provide all necessary feature keys as defined by the model. Check the model's documentation or the library's source for required keys (e.g., in the `__init__` or `predict` method definitions).
```
ModuleNotFoundError: No module named 'webrtc_models'
```
cause The `webrtc-models` library is not installed in the current Python environment.

fix Install the library using `pip install webrtc-models`. Ensure your Python environment is correctly activated.
```
ImportError: cannot import name 'models' from 'webrtc_models'
```
cause Attempting to import an internal module or a non-existent symbol directly from the top-level `webrtc_models` package.

fix Import specific model classes directly from the `webrtc_models` package (e.g., `from webrtc_models import PacketLossPredictor`). Do not try to import submodules like `webrtc_models.models`.

Warnings

breaking The `predict` method for all models (e.g., `PacketLossPredictor`, `BandwidthPredictor`, `QoEPredictor`) changed its signature in version 0.3.0.
Fix: Update calls to `predict` to pass a single dictionary of features instead of individual keyword arguments. For example, change `predictor.predict(arg1=val1, arg2=val2)` to `predictor.predict({'arg1': val1, 'arg2': val2})`.
gotcha Models expect specific feature keys in the input dictionary passed to the `predict` method. Missing or misspelled keys will raise a `KeyError`.
Fix: Refer to the model's documentation or the library's source code for the exact feature keys required (e.g., `inbound_rtp_packets_lost`, `inbound_rtp_packets_received` for `PacketLossPredictor`).
gotcha The `webrtc-models` library explicitly requires Python 3.12 or newer.
Fix: Ensure your environment is running Python 3.12.0 or higher. Use `python --version` to check and upgrade if necessary, or activate a suitable virtual environment.

Install

pip install webrtc-models Install latest version

Imports

PacketLossPredictor

from webrtc_models import PacketLossPredictor

BandwidthPredictor

from webrtc_models import BandwidthPredictor

QoEPredictor
```
from webrtc_models import QoEPredictor
```

Quickstart

Demonstrates how to import a model, initialize it, provide input features as a dictionary, and get a prediction.

from webrtc_models import PacketLossPredictor

# Initialize the predictor model
predictor = PacketLossPredictor()

# Prepare features as a dictionary
# (Note: Specific keys are required by each model)
features = {
    'inbound_rtp_packets_lost': 0,
    'inbound_rtp_packets_received': 100
}

# Make a prediction
prediction = predictor.predict(features)

print(f"Predicted packet loss: {prediction}")

view raw JSON →