Crop Yield Estimation

Inspiration

AI Hack

What it does

Estimates Crop yield for counties in the state of Illinois

How we built it

Python (Pandas and Sklearn), Rapidminer. Mean features generated from Temperature and EVI

Challenges we ran into

Incomplete and unnecessary data, (i.e. temperature values missing when EVI is available). Data is not aligned, we needed to estimate the closest data points and match 2 data sets.

Accomplishments that we're proud of

RMSE is 12 when the mean yield is 185 proving our model is useful.

What we learned

Data mining, feature generation, the importance of understanding the subject of study. Training the model on data outside 2015-2019 decreases the accuracy, also the data points should be taken within the vegetation region Apr-Nov.

What's next for Crop Yield Estimation

Use more diverse data and more features. Generate min/max features from EVI and temperatures. Use VOD, soil moisture data. Use neural networks (LSTM).

Built With

• pandas
• python
• sklearn