Machine Learning -Water level predictor

water level for post - monsoon (2018)
The idea

Inspiration

India is backed up by agriculture and Ground water becomes a necessary source for irrigation. That's why we decided to step up and do our job to help agriculture and as well the environment.

What it does

The decadal average of the pre and post monsoon groundwater level is compared with the present situation. The villages,blocks,districts and states showing improvement / decline with time ( one decade ) are located The random forest regression prediction model is used to categorize the trend of the improvement / decline in terms of villages,blocks,districts and states with high improved accuracy. Based on the predictions, the improvement / declining trend, critical zones are identified and represented in maps

How we built it

We used Jupyter Kernel to train and test our models (using scikit-learn) and geoplot, geopandas to plot the values to a map.

Challenges we ran into

The dataset we had was limited to a particular state (TamilNadu) and the dataset itself had a lot of anamolies and NaN values. This made preprocessing difficult.