Machine learning pilot project by Microsoft helps in improving crop yield by 30% in Andhra Pradesh
Following the launch of the pilot in June 2016, that tested a new Sowing Application for farmers combined with a Personalized Village Advisory Dashboard for the Indian state of Andhra Pradesh, the results show a 30% higher average in yield per hectare. The Sowing App was developed to help farmers achieve optimal harvests by advising on the best time to sow depending on weather conditions, soil and other indicators. The pilot was implemented in Devanakonda Mandal in Kurnool district and the Advisory applied only to the groundnut crop.
This was made possible through a partnership between the International Crops Research Institute for the Semi-Arid Tropics (ICRISAT), Microsoft, and the Andhra Pradesh government. The pioneering digital tools were released as part of the Andhra Pradesh Primary Sector Mission (APPSM) also known as Rythu Kosam. The results of the pilot were announced today at a panel discussion at the 20th National Conference on e-Governance taking place in Visakhapatnam.
ICRISAT adopted Microsoft Cortana Intelligence Suite including Machine Learning (ability of computers to learn without being specifically programmed) and Power BI or Business Intelligence, to empower farmers and government officials with technology, and promote digital farming practices in the state. The Personalized Village Advisory Dashboard has been especially developed to enable officials of APPSM to better manage programs of scale. Using powerful BI tools, this dashboard provides important insights around soil health, fertilizer recommendations, and seven-day weather forecasts derived from the world’s best available weather observations systems and global forecast models. This data is then downscaled for the highest possible accuracy at the village level, to transform how small holder farmers tackle climate change to drive effective decision-making for their crops.
During the pilot, farmers were sent 10 sowing advisories starting June 10 via SMS in Telugu comprising essential information such as sowing recommendation, seed treatment, optimum sowing depth, preventive weed management, land preparation, farm yard manure application, recommendation on harvesting, shade drying of harvested pods, and storage etc. Going forward, when combined with other data collected from the Rythu Kosam project, rich datasets can be processed to build other predictive models for the farmers.
Commenting on this development, Anil Bhansali, Managing Director, Microsoft India (R&D) Pvt. Ltd., said, “This result is proof of the efficiencies that cloud technology can bring and thereby help farmers economically by improving agricultural productivity. The applicability of the Intelligent cloud is a significant start for digital agriculture and we hope more governments and stakeholders come forward to reap benefits of this innovation.”
The Sowing Application utilized powerful artificial intelligence to interface with weather forecasting models provided by USA based aWhere Inc. and extensive data including rainfall over the last 45 years as well as 10 years of groundnut sowing progress data for Kurnool district. This data was then downscaled to build predictability and guide farmers to pick the ideal sowing week.
C Madhusudhana, President, Chaitanya Youth Association and Watershed Community Association of Devanakonda said, “Registered farmers did the sowing in the last week of June and managed the crop as per the weather-based advisories provided by ICRISAT and Microsoft throughout the crop-growing period. Farmers who sowed in the first week of June got meagre yields due to a long dry spell in August; while registered farmers who sowed in the last week of June and the 1st week of July and followed advisories got better yields and are out of loss. Many more farmers are showing interest to register their mobile phone numbers for receiving the advisories.“
G Chinnavenkateswarlu, a farmer from Bairavanikunta village said, “I have three acres of land and sowed groundnut on June 25, 2016, based on the sowing recommendations provided to me by ICRISAT. My crops were harvested on October 28, 2016 and the yield was about 1.35 tonne per hectare. Advisories provided for land preparation, sowing, and need-based plant protection proved to be very useful to me.”
“We are excited about the results that have emerged from the use of the Sowing Application and Personalized Village Advisory Dashboard. We look forward to continuing our partnership with Microsoft to enhance incomes and improve the lives of small holder farmers, and give a boost to our digital agriculture initiative in a big way,” said Dr. David Bergvinson, Director General of ICRISAT.
ICRISAT is providing technical backend support to Rythu Kosam, which is aimed at positioning the state among the best three performing states by 2022. This technical input involves the establishment of pilot sites of learning in 13 districts of 10,000 ha each; upgrading soil analysis laboratories; technical support for planning; as well as adopting an Inclusive Market-Oriented Development (IMOD) strategy to benefit smallholder farmers through public-private partnerships; and promoting private investments in the state.
