water stress proj

This project will be inspecting the effects of water stress found from remote sensing data.

Indentifying points of interest using change point detection

Use Casual Impact to measure the effect of a point of interest

Applying the first two point to forcasts to aid preventative measures

How to Use

1. Load and normalize data

input data should be monthly resampled dataframe with NDVI,SPEI and predictive/correlating variables to help the forecasting

2. Train model

3. add forecasted data to original data

4. init bench testing

5. run benchtesting

6. calculate casual impact for changepoints

first casual impact from dataframe is shown

File descriptions:

filename	description
segmentation1.ipynb	Creating the synthetic dataset
---	Benchtesting the algorithms on the synthetic dataset
segmentation2.ipynb	prototyping on the real NDVI dataset
---	check witch changepoints like in the drought-linked pointer years
all_in_one___periodic_wave.ipynb	an example on how to use the classes made in this repo to benchmark the change point algorithms on the datasets
benchmark.py	benchmarking class to run the benchmarks
helper.py	file containing helper functions
lstm_model.py	file containing the model to facilitate forecasting
r_files folder	contains all the necessary files to run the R environment to do the checks for BFAST and BEAST on the synthetic dataset

requirements

cpdetect: https://github.com/jzyee/cpdetect
ruptures: https://centre-borelli.github.io/ruptures-docs/
casualimpact: https://github.com/dafiti/causalimpact

References

Ensign DL and Pande VS. Bayesian Detection of Intensity Changes in Single Molecule and Molecular Dynamics Trajectories. J. Phys. Chem B 114:280 (2010)
C. Truong, L. Oudre, and N. Vayatis. Selective review of offline change point detection methods. Signal Processing, 167:107299, 2020.

jzyee / water_stress_proj