After four years of supporting Cambodia's National Institute of Statistics on flagship exercises, including the annual Cambodia Agriculture Survey and the country's decennial census, one thing became clear to me: the government is strong at producing data, but there is a significant untapped opportunity in how that data is used.
Cambodia's statistical system generates rich, nationally representative datasets year after year. The infrastructure, the fieldwork capacity, and the institutional commitment are all there. What has been less developed is the analytical layer that sits on top, the methods that can turn raw survey data into predictive insights, automated workflows, and evidence that directly informs policy.
That recognition is what led to this training programme. Under the FAO's Technical Cooperation Programme framework, the Royal Government of Cambodia requested technical assistance to build machine learning and advanced analytics capacity within NIS and the Ministry of Agriculture, Forestry and Fisheries. Having seen the need firsthand through years of working alongside these institutions, I was glad the request came through and eager to support it.
Why This Matters Now
We are living through a period of rapid growth in artificial intelligence. Large language models are reshaping how organisations process text and extract knowledge. Machine learning is being applied to everything from crop yield forecasting to poverty mapping. Geospatial and climate data are more accessible than ever through open APIs. Governments that understand these tools and can apply them critically will be better positioned to design responsive, evidence based policy. Those that don't risk falling behind, not in data collection, but in data use.
The goal is to expand the toolkit available to statisticians and analysts so they can ask bigger questions of the data they already collect.
What the Training Covers
The programme is designed as a crosscutting exercise spanning more than two weeks. Rather than a narrow focus on a single technique, it walks participants through the full landscape of modern data science as it applies to official statistics. Topics include large language models and how they work, supervised machine learning for prediction tasks, connecting to geospatial and climate data through APIs such as Google Earth Engine and NASA POWER, feature engineering from survey and remote sensing data, and model evaluation and interpretation.
Everything is built around Cambodia's own data: real agricultural survey records, real climate variables, real geographic boundaries. The goal is not abstract understanding but practical capability that participants can carry forward into their own work.
A Foundation for What Comes Next
This training is one piece of a broader effort. Other international partners are also planning advanced analytics training for many of the same staff later in the year, using the same technical environment we establish here. The intent is cumulative: build a shared foundation in Python, Jupyter, and cloud computing, then layer progressively more advanced methods on top.
If governments are going to make the most of the data they work so hard to collect, investments like these are essential. I am proud to be part of that effort in Cambodia.