Internship Experience:

Statistical Scientist Intern, Bayer Crop Science R&D, Chesterfield, MO                                                                       

Feb 2023 – Jun 2023

• Used GMM, Spectral Clustering, BIRCH and K-Means to determine high yielding soil environmental clusters
• Employed PCA and MDS to select important features per soil environmental cluster based on holding values
• Applied MARS, BORUTA, Step-wise regression, XGBoost and Random Forest to rank the soil parameters instrumental in driving yield
• Explored the difference in yield of genetically modified crops by studying the interaction between soil clusters and treatments using Linear Mixed Models
• Identified the characteristics of the non-performing clusters and gave statistical recommendations to be applied in future

 

Research Intern, Texas A&M AgriLife Research and Extension Center, Corpus Christi, TX                                     

May 2022 – Aug 2022

• Formulated training datasets using K-Means clustering and Dynamic Time Warping (DTW) techniques of data modeling for com­bining discontinuous non-stationary time-series data from cultivation years with varying growth period
• Used Transfer Learning approaches and Frequency Domain transform to predict early season growth indices
• Designed a variety of LSTM models like Stacked LSTM, Bi-Directional LSTM, CNN LSTM, Convolutional LSTM and Encoder-Decoder LSTMs to predict in-season Canopy Cover, Canopy Height and Excessive Green Indices of cotton crops
• Developed optimal ARIMA models using inferences drawn from ACF/PACF plots to predict the canopy features until the end of the cultivation period using a pipeline of methods to remove non-stationarity
• Predicted the yield with 91% aggregate accuracy for cotton, wheat and forage crops using Multi-Layer Deep Learning models