I am Arunit, owner of Arunit Farms, an agriculture start-up where we grow fruits and vegetables hydroponically, independent of climatic conditions. The project has been in the running for 4 years now. PGP-DSBA has provided a perspective on the importance of data in our life, and we have used that perspective to start collecting data on the different plant cycles running in our project.
There is a variation in the yields of the various crops we have planted over the years, which has never been understood till now. Some cycles have seen crops with outstanding performance, whereas others have crops with average and poor performance, and other parameters have remained constant. The problem involved identifying the best parameters for consistent yield for the different crop cycles with minimal fertilizer and electricity wastage.
The information collected involves the water temperature, air temperature, humidity, insolation, the quantity of fertilizer used, quality/brand of fertilizer used, electricity consumed in maintaining the air temperature, plant growth in a specific period, and yield per plant.
We performed this exercise on English Cucumbers, cherry tomatoes, and red/yellow capsicum, where different air and water temperatures and fertilization schedules/rates were maintained on different samples.
We used Data Science to solve this problem by building linear regression models to predict which features affected the yield the most and the coefficients by which increments and reductions in the future values affected the yield.
The model was deployed on the next crop cycle, which saw a significant boost of around 12% in the yields of the three crops. In contrast, electricity consumption was reduced by 4% approximately. The fertilizer consumption remained almost the same, but certain brands of fertilizers gave better results.
