Jnsll / LAPrediction

Loop Aggregation Prediction

Context-Aware Approximate Scientific Computing

This repository is a companion piece to the article entitle "Context-Aware Approximate Scientific Computing". It provides the code used for collecting the data (as well as applying the loop aggregation technique), in folder data_collection, and the code used for the evaluation in the evaluation folder.

Requirements

Data collection

• Python3
• flopy==3.3.5
• pandas==1.4.2
• matplotlib==3.5.1
• scipy==1.8.0
• gdal==3.4.0
• numpy==1.22.3
• jupyter-notebook
• libgfortran5

Evaluation

• pandas
• numpy
• sklearn

Usage

To directly go to the results of the evaluation process, we recommend to open the corresponding jupyter notebooks in folder evaluation/notebooks:

• CostPrediction.ipynb for the evaluation of the cost predictive model (RQ 1.1)
• ValidityPrediction.ipynb for the evaluation of the validity predictive model (RQ 1.2)
• Approach.ipynb for the evaluation of the overall approach (RQs 2 & 3 & 4)

For the data collection, we recommend going to the provided example in data_collection/example. The necessary information to run it is given in the associated README.md.