Acea-Water-Challenge

In this project I used the dataset by Acea about the Doganella Aquifer in Italy and predicted the depth to groundwater by using a Random Forest. The notebook contains my analysis and implementation of my Random Forest model. It is divided into sections for data exploration, data cleaning, statistical tests, feature engineering, and modeling.

Packages Used

• seaborn==0.10.1
• numpy==1.18.5
• pandas==1.0.5
• matplotlib==3.2.2
• statsmodels==0.11.1
• scikit_learn==0.24.1

Approach

1. Explored the data to find any patterns or points of interest.
2. Cleaned the data using interpolation and selecting a time range which would yield the most accurate entries.
3. Tested for stationarity and trend by using the ADF and KPSS statistical tests.
4. Implemented walk-forward validation with the Random Forest.

Results