I recently had the opportunity to work on a very neat data science project. We used historical data on weather, soil composition, and crop yields from a dozen farms over a decade and employed a number of regression and classification methods (linear, tree-based, and neural nets) to predict future yield.
Coding was done in Python using Jupyter Notebooks and the following packages:
NumPy, Pandas, GeoPandas, GDAL, MatPlotLib, Pillow, and Scikit-Learn
This gave me the opportunity to not only learn some fun packages but to also communicate regularly with some very smart folks doing important agronomy work advising farmers on fertilizer treatment plans. I learned some feature reduction and hyperparameter tuning methods, and in general learned a lot about the process of working with data for good.
I am also interested in applications of genetic algorithms. I recently completed a collaborative project using genetic algorithms to predict game values of impartial combinatorial games.
