File Entry: Machine Learning for Soil Fertility and Plant Nutrient Management using Back Propagation Neural Networks

The  objective  of  this  paper  is  to  analysis  of  main  soil  properties  such  as  organic  matter,  essential  plant  nutrients, micronutrient  that  affects  the  growth  of  crops  and  find  out  the  suitable  relationship  percentage  among  those  properties  using Supervised Learning, Back Propagation Neural Network. Although these parameters can be measured directly, their measurement is difficult and expensive. Back Propagation Networks(BPN) are trained with reference crops’ growth properties available nutrient status  and its ability to provide nutrients out of its own reserves   and through external applications for crop production  in both cases,  BPN  will  find  and  suggest  the  correct  correlation  percentage  among  those  properties.  This  machine  learning  system  is divided into three  steps,  first  sampling  (Different soil  with same  number of properties with different p arameters)  second Back Propagation  Algorithm  and  third  Weight  updating.  The  performance  of  the  Back  Propagation  Neural  network  model  will  be evaluated using a test data set. Results will show that artificial neural network with certain number of   neurons in hidden layer had better performance in predicting soil properties than multivariate regression. In conclusion, the result of this study showed that training is very important in increasing the model accuracy of one region   and  result in the form of a guide to recognizing soil properties relevant to plant growth and protection.