Data Scientist

The Statistics Division (ESS) develops and advocates for the implementation of methodologies and standards for data collection, validation, processing and analysis of food and agriculture statistics. In these statistical domains, it also plays a vital role in the compilation, processing and dissemination of internationally comparable data, and provides essential capacity building support to member countries. In addition, the Division disseminates many publications, working papers and statistical yearbooks, which cover agricultural and food security relevant statistics (including prices, production, trade and agri-environmental statistical data). The Statistics Division is involved in the management of a number of large scale projects, aimed at improving statistical methodologies and establish best practices for the collection, collation, processing, dissemination and use of data relevant to food security, agriculture and rural areas.

Within the Statistics Division, a Data Lab for Statistical Innovation has recently been established to lead the Division’s work related to data science applications (text mining, machine learning, web scraping, geospatial analysis) and the use of big data to solve real data problems in agriculture statistics and policy analysis. In particular, the Data Lab will support the Hand-in-Hand Initiative aiming at accelerating progress on Sustainable Development Goals in 43 focus countries with the highest poverty and hunger rates.

Rapid technological development requires ESS to innovate in a variety of areas to modernize the statistical business process and meet the increasingly demanding needs for fast, accurate, easy and cost-effective data. Therefore, it is part of FAO’s strategy to engage with non-official, Big Data sources including geospatial data and to rely on data science methods to solve the current data gaps problems.

The Data Lab initial objectives are (i) to promote the use of non-official, unstructured data and data science methods to fill data gaps in domains and geographical areas in which there is little official data available. In the context of the Hand-in-Hand Initiative, it will focus on building data systems that will facilitate the identification of target areas and will describe different aspects of their agricultural potential; (ii) to promote the use of non-official, unstructured data and data science methods to validate official data reported by countries, to identify areas of future collaboration and technical assistance and (iii) to build tools to extract data from texts in order to summarize available information that characterize the situation of specific territories and to map/identify/categorize effective policy interventions that can be applied in similar situations.

The final objective is to increase the quantity, quality and timeliness of the statistics that the Organization produces to serve the countries and help achieving the Sustainable Development Goals.

The data scientist, whether Consultants and PSA subscriber, will join this new Data Lab.

Reporting Lines

Consultants and PSA subscribers in ESS will work under the immediate supervision of the Methodological Innovation Team Leader and the general oversight of the Director and the Deputy Director of the Division. They may be called upon to collaborate with the IT division (CIO), the Office of the Chief Statistician (OCS) and relevant technical divisions.

Consultants and PSA subscribers will focus on one or more of the following technical areas\:

• Advanced programming with statistical tools, languages and ML libraries such as R, Python, SQL and Spark ML, Hadoop environment.
• Advanced programming with both open and commercial geospatial software (e.g. Google Earth Engine, QGis, ESRI).
• ETL development, Big Data storage.
• Text mining and analysis with Natural Language Processing\: word categorization and tagging, syntactic parsing, word sense disambiguation, topic modeling; contextual text mining, application of machine learning to NLP, semantic similarity, phrasal semantic analysis, text matching and similarity, word embedding, lexicon normalization, named entity recognition. Experience with NLP libraries such as NLTK, openNLP, Stanford-NLP, WordNet or other NLP software.
• Data mining and analytics.
• Data visualization and presentation, turning complex analysis into insight. Familiarity with leading visualization tools (e.g., Shiny Apps, Power BI, Tableau and d3.js)

The primary responsibility of this position is to employ big data, geospatial data and data science techniques to develop a range of solutions. In one or more of the above mentioned domains, Consultants and PSA subscribers will contribute to and/or take responsibility for one or more of the following tasks\:

• Identify suitable big data sources or EO sources where appropriate;
• Develop solutions using big data such as machine learning and statistical techniques to improve current estimates of statistical variables (e.g. agricultural prices, crop acreage and yields, land use,..) in terms of their coverage, timeliness and accuracy;
• Develop solutions using text mining methods such as clustering, classification, information retrieval, topic discovery, summarization, topic extraction;
• Develop models for semantic analysis using natural language processing (NLP) and analytical methods that are accurate and scalable;
• Be proficient in quickly developing models, as well as taking prototypes to full production;
• Perform operationally-relevant data analysis ranging from text mining, to web scraping, machine learning, crowdsourcing, around structure and unstructured data sources; develop and apply new and existing models and algorithms;
• Perform operationally-relevant data analysis ranging from image processing, to geospatial analysis and machine learning methods; develop and apply new and existing models and algorithms;
• Publish well documented, reproducible work;
• Assist in developing forecasting and nowcasting models using data from social networks, on-line databases, administrative data, satellite imagery or transaction records.

They will perform other related duties as required.