Would Better Data Visualisation Improve Farm Profitability?
News Sep 21, 2015
“Understanding the cost of production and relating this to the return on increased yield is very complex and the current benchmarking tools are of limited value,” Ward says.
His comments are supported by research by Mark Reader of the Department of Land Economy at the University of Cambridge, one of the co-chairs of the SIG. In a recently released paper, ‘Loss-making marginal spending on crop variable inputs’, data from the Farm Business Survey 2004-2012 was used to assess the crop gross margins and input spending for conventional winter wheat and oilseed in England and Wales.
It concluded that although additional fertiliser increased yield, the economic value of this yield in many cases was less than the cost of the inputs. Reader concluded that unknowns such as yield, quality and price make it very difficult to estimate the economic optima for inputs.
These conclusions hold across a wide range of alternative economic models and subsets of the data.
Ward says that currently farmers benchmark themselves against comparable farms, ie those of a similar size and crop portfolio, but it is difficult to drill down to see if additional expenditure results in a profitable crop.
“If you compared all the wheat farmers in the UK, some use £80 of fertiliser per hectare and others £180; that’s a massive spread. If you are using £130, do you need to increase or decrease the amount? A farm using £180 per hectare might be more profitable but is this related to the inputs? It could be better soil, less disease, better weather, higher price.”
“Therefore, I think this method is fundamentally flawed: it doesn’t tell you which inputs are related to profit.”
Improved data visualisation would help farmers to benchmark against their own data; keeping records over a period of time would allow a better helicopter view of where costs were incurred and revenue generated.
Precision farming has dramatically increased the amount of data available, from mini-weather stations giving field conditions, drones visualising crop health and telemetrics on machinery utilisation and fuel costs.
Matthew Smith, computational ecologist at Microsoft Research and co-chair of the SIG, says:
“There are so many variables in agriculture, knowing as much as you can about each of them means you are in better position to create models to help informed decision making. However, you need a system in place that can extract meaningful information from this data. We need the input of farmers to determine what metrics they would find most insightful.”
Eric Hannell, Senior Project Consultant at Tableau, agrees. His particular interest is ‘dashboards’ which provide a snapshot of what is happening, hiding the complexity so that good decisions can be made.
“Even the perfect dashboard will only answer the questions you designed it to answer, so farmers need tools that they are able to tweak themselves so that they are tailored to their requirements. A good dashboard should be able to adapt to changes in needs and provide ad hoc visual analysis of the data.”
After the presentations there will be an opportunity for participants to discuss the issues in more depth.
Agri-Tech East’s ‘Taking Data Presentation Seriously’ Big Data Special Interest Group is being held at Rothamsted Research on the 29th September, 2015.
MIT researchers have developed a cryptographic system that could help neural networks identify promising drug candidates in massive pharmacological datasets, while keeping the data private. Secure computation done at such a massive scale could enable broad pooling of sensitive pharmacological data for predictive drug discovery.