Skip to content

Latest commit

 

History

History
38 lines (32 loc) · 2.9 KB

README.md

File metadata and controls

38 lines (32 loc) · 2.9 KB

Project 2 - Predicting Ions Concentration in Water Streams

Names : Berezantev Mihaela, Sinsoillier Mike Junior, Du Couédic De Kergoualer Sophie Zhuo Ran

Background

This repository contains the code to produce two models dedicated to predict the ion concentrations in water streams using data measures from in-situ probes.

This repository contains:

Prequisites

The notebooks can be executed as they are in Google Colaboratory. If you want to run them locally, make sure to have the following packages installed.

  • python3. All the implementation are coded in python3.
  • pandas. Used for data exploration and preprocessing.
  • pytorch were used to compute the fast Fourier transforms during features creation.
  • numpy : for an easy manipulations of the data arrays. You can install it via pip.
pip install numpy
  • scikit-learn is a simple library for machine learning. Used both in the RNN and boosting regressor.
  • TensorFlow
  • keras is a deep learning API running with TensorFlow. It is necessary to run for the RNN.

Usage

To run and produce the results made by the boosting regressor or the RNN, a folder with the path /content/drive/MyDrive/ML/Project 2/data should be created in google drive containing the two .csv files Erlenbach_ion_concentration.csv and Erlenbach_probe_data10min.csv. The two notebooks can then simply be executed in Google Colaboratory.

If runned locally. The notebook should be updated with the appropriate paths to the data files, and the first cells of each notebook removed:

from google.colab import drive
drive.mount('/content/drive')