Estimated Time of Arrival

A simple machine learning system designed to estimate how much time a task will take

Running the Demo

System Requirements:

Using the default options the demo takes about 1-2 minutes on a single core (2.8GHz max clock) using roughly 1.5GB of RAM

pip install requirements.txt
python demo.py

Demo.py accepts three arguments:

1. nodes The number of nodes each synthetic script should have, default: 10000
2. scripts The number of synthetic scripts you want to generate, default: 10-100
3. nodeTypes A list of different node types you want to generate, default: nodeTypeA nodeTypeB nodeTypeC nodeTypeD nodeTypeE nodeTypeF

Usage

Parser

1. Subclass Script for your relevant DCC
2. Initialise a FileStorage object (or your own Storage subclass) with a file path
3. Initialise Subclass using the Initialised Storage Object
4. Feed subclass.parse into subclass.write
5. Repeat for all scripts

Interpolator

1. model = EvaluationModel(model=BayesianRidge) Initialise the EvaluationModel, providing your preferred Algorithm
2. model.fit(scripts_with_known_render_times, execution_times) Fit the parsed script data to the execution time data
3. prediction = model.predict(scripts_with_unknown_execution_times) Make a prediction on scripts where we don't know the execution time already
4. Save out your trained model for future use