Sub-Theme Food Safety (MARDI)

Food safety surveillance in Malaysia is capacity-limited. Pesticide residue testing, heavy-metal screening, and farm inspection are laboratory-intensive and expensive, so only a small fraction of what is grown can ever be sampled. Sampling is therefore typically allocated by fixed rotation, by commodity importance, or in response to a complaint, that is, reactively.

Meanwhile, the agronomic conditions that raise chemical-input pressure on a crop are already recorded, annually and free of charge, in MARDI's crop statistics. Planted area, harvested area, production tonnage, and farm-gate value are published for every crop in Pahang from 2015 to 2024. Nobody is currently using these as an early-warning signal for where residue risk is concentrating.

Analysis of MARDI's Pahang crop statistics (2015–2024) reveals a clear intensification signature in the state's vegetable sector: planted area is expanding rapidly while yield per harvested hectare is falling.

The problem is that none of this information reaches the people deciding where to sample. A crop showing a decade of yield erosion under expanding area receives no more surveillance attention than a stable one. MARDI and food-safety authorities have the data to form a risk prior, but no instrument that converts it into a sampling priority.

1. To construct, from MARDI crop statistics, a set of agronomic stress indicators (yield trend, area expansion rate, harvest-failure ratio, and price volatility) for every crop grown in Pahang.
2. To identify which crops exhibit the intensification signature (expanding area with declining yield) and quantify the strength of that signal.
3. To develop a Crop Surveillance Priority Index (CSPI) that ranks crop-year combinations by estimated food-safety risk exposure, to direct limited sampling capacity.
4. To detect and flag data-quality defects in the source statistics that would otherwise distort any downstream analysis.
5. To deliver a dashboard allowing MARDI officers to see, for any year, which crops the model ranks as highest priority and why.

Primary: MARDI Pahang crop statistics 2015–2024 covering fruits (45 crops), vegetables (56 crops), and cash crops (7 crops), each with planted area, harvested area, production, and farm-gate value per year. This yields approximately 978 crop-year records.

• A cleaned, machine-readable, long-format version of MARDI's Pahang crop dataset, reusable by other researchers, since the published workbook is not analysis-ready.
• A documented data-quality report identifying imputed values, missing-value patterns, and inconsistencies in the source.
• A ranked Crop Surveillance Priority Index for all major Pahang crops, updated annually as MARDI publishes new figures.

• Yield decline has causes other than intensification, including varietal change, weather, and shifting land quality as cultivation expands onto marginal land. The analysis should acknowledge these competing explanations rather than assert soil fatigue.
• Reported harvest ratios appear imputed for several vegetables, limiting the reliability of failure-rate features for those crops.
• The data is state-aggregated, so district-level targeting requires joining to DOSM's district crop datasets, whose crop categories may not align exactly with MARDI's.