Data Driven Yield Predictive Analytics of Major Crops in Punjab

Abstract

The economy of Pakistan is primarily dependent on the production of agricultural and agro-based products. Many factors including climatic conditions, soil fertility, topography and water quality reduce agricultural productivity that leads a threat to food security and financial loss to farmers. Many factors affect crop yield, including rainfall, temperature, humidity, soil, and PH. These factors pose significant risk to farms leading to yield reduction when they are not properly monitored and managed. To ensure professionalism, transparency, and integrity, Pakistani regulatory authorities like the Bureau of Statistics (BOS) and the Meteorological Department provide accurate, relevant, timely, and user-friendly data of crops. In this research, two different time series datasets were arranged in systematic order, consisting of crop production (Crop type, district, area, yield) and environmental factors (rainfall, temperature). The aim was to analyze the large collected of data of major crops in south Punjab (Wheat, Rice, Cotton) to identify patterns in different locations and extract main features that were used to predict the yield of crops. Farmers could use this approach to determine which crops are most suitable for their location on the basis of historical data. In the proposed study, data mining techniques were used for processing data, identifying patterns and extracting useful features to predict crop yields using these features.