The problem is taken from one of the Kaggle machine learning competitions, where they exposed the datasets with a bit of information and asked the contestants to make a prediction from a test dataset using machine learning techniques. It is about the Africa Soil Property Prediction Challenge. The training dataset contains different measurements of the soil and it is expected that the contestants are able to predict values for five properties of the soil (target variables): SOC, pH, Ca, P, and Sand. Full details regarding this can be found at https://www.kaggle.com/c/afsis-soil-properties/.
For the purpose of this case study, we will choose just one target variable named P and ignore the others. Once you are able to predict one target variable, you can follow the same approach and predict others. So, you should try predicting all the variables by yourself.
