Grow Asia Hackathon 2018
Leveraging data to help smallholder farmers thrive

Data

Dataset Guide

The following dataset has been prepared to help you create, test and illustrate your solution. It is made up of seven tables and two APIs.


Growasia datasets

 

Is the data real?

The mill, price, weather and satellite data is real data. The farmer, shipment, warehouse and retailer data is a projection prepared just for the hackathon. Projected data has been chosen to reflect the reality in Sumatra as closely as possible.

What can we do if data is missing?

Creativity is encouraged. If data is missing from the set (such as the farmer’s Facebook ID) but is important to your solution, include it. If you are working with missing data, you need to show how your solution will either secure or generate this data.


 
 

Ownership

The data in the table is owned by different actors such as the public, mill, end buyer or is available under a commercial license:

  • Public Data – you can freely use public data to devise your solution.
     
  • Commercial Data – If you use commercial data, you will need to demonstrate how you will cover this cost. Please note that some satellite and climate data is available for free.
     
  • End Buyer and Mill – If you use this data you will need to show how you plan to negotiate its release with the owner. Your solution may also generate this data internally.

What correlations will you find in the data?


A correlation between the productivity of a farm and its distance to the nearest mill, fertiliser usage, use of crop protection products (CPPs), tree age, training attendance, farmer education, land ownership type, plant variety, soil grade, and collection frequency. Production will also vary from month to month reflecting rainfall as well as with crude oil price (CPO).  You will note that a disease outbreak occurs in April and May 2016 which impacts production.

What if a correlation is missing?

If the pre-reading documents highlight a correlation (eg. road grade and yield) which is not present in the data you can still use this correlation to drive your solution. For more information please join us at our Pre-Hackathon Workshop where we will be presenting on the available datasets.


Mill Locations

 

Click here if the map does not load.


Technical Partner API


 
 

Dataset Glossary

Smallholder - A smallholder will typically farm about 1 to 10 hectares of land. Most of the labour on the plantation will be undertaken by family members. They will often grow a crop (like palm oil) for cash and tend a garden and livestock for their own consumption. The pre-reads including one document on palm oil smallholder livelihoods.

Fresh Fruit Bunch (FFB) - The fresh palm fruit grows in a bunch, weighing about 10 to 40 kg. The fruit grows up to 15 metres from the ground. Significant effort is need to cut it from the palm, and transport it by wheelbarrow to a truck loading point. Ensuring the fruit reaches a mill within 24 hours is important to maximuming the quality of and quantity of oil which can be extracted.

Crude Palm Oil (CPO) - Crude Palm Oil is the primary product produced from the FFB, and is further refined for use in foods, cosmetics and industrial applications.

Crop Protection Products (CPP) - Include herbicides, insecticides and fungicides. Selected and used correctly, they reduce the likelihood and impact of pests and diseases.

Oil Palm Mill - The mill processes the Fresh Fruit Bunches (FFB) into Crude Palm Oil (CPO) as well a kernel oil.

Retailer - The retailers are a general agriculture store, they sell crop protection products and fertilisers to the farmers.

Warehouse - Crop protection products and fertilisers are held at warehouse by the manufacturer, for on-sale to retailers.

Soil Grade - A range attributes of the plantation’s soil affect palm oil yields including consistency and nutrients. In the dataset, the soil is graded from 5 (best) to 1(worst).

Planting Date - The year the plantation was planted. The palm’s productivity varies substantially over its lifetime of up to 30 years.

End buyer - The end buyer of palm oil is in many cases a Fast Moving Consumer Goods (FMCG) company. Many FMCG companies have made a commitment to palm oil traceability and sustainability. Unilever’s palm oil sustainability policy is included in the pre-readings.

Yields - A single palm oil palm will produce about 30 to 60kg of FFB each year. The pre-readings include three documents which outline the key drivers of yield.

Planting Density - The number of palms per hectare of land.