masterdoors / soil_temp

soil_temp

Main files and folders:

--data - dataset folder

--test_outputs - experiment log folder

--test_outputs/build_table.ipynb - a notebook to build the experiment result table.

--boosted_forest.py - Scikit-learn-based (slow) implementation of Boosted Forest --soil_temp_experiments.ipynb - experiments on the soil temperature dataset --standard_datasets.ipynb - experiments on standard datasets --test_hidden_size.ipynb - test size of the hidden layer on the Diabetes dataset

Setup

1. Download COSMOS-UK dataset via the notebook https://github.com/masterdoors/soil_temp/blob/main/data/UK/get_dataset.ipynb. If the scipt does not work, please download the dataset manually via the link: https://catalogue.ceh.ac.uk/documents/399ed9b1-bf59-4d85-9832-ee4d29f49bfb and put all the data to the folder: data/UK/catalogue.ceh.ac.uk/datastore/eidchub/

2. Install requirements for the experiments with boosting and ARIMA:

   2.1. Install python 3.9

   2.2. Install dependencies:

   pip install -r requirements_boosting.txt
   

   2.3. Run the experiments (soil_temp_experiments.ipynb, standard_datasets.ipynb)

3. Install requirements for the experiments with neural networks:

   3.1. Install python 3.11

   3.2. Install dependencies:

   pip install -r requirements_tkan.txt
   

   3.3. Run the experiments (soil_temp_experiments.ipynb).