Africa Soil Property Prediction and Clustering

Objective

The objective of course project was to go through all the steps in data science methodology such as problem definition, data collection, data pre-processing/cleaning, exploratory data analysis, feature engineering, model development, hyper-parameter tuning.

Problem Definintion:

Measuring soil properties such as Ca, P, pH, SOC using tradional laboratory based methods is expensive. We developed a regression model to predict these proprties using cheap spectroscopy data.

• Kaggle Challenge
• project report
• presentation
• github repo

Tools & Technologies

Regression, Clusterng, Machine Learning, PCA, Pandas, SkLearn, Ensemble