Food Consumption data decomposition

[rbroth]

When importing food consumption data, e.g. from IHS4 (Fourth Integrated Household Survey), their classification of foods will not be in a format useful for our automated matching to Food Composition Table entries. For example, the survey could report food in terms of dishes consumed ("banana snack", "beef stew").

In such a case, the data needs to be decomposed into constituent parts that have a corresponding entry in the Food Genus list ("banana snack" --> "50% banana, 30% white sugar, 20% wheat flour"; "beef stew" --> "50% beef, 35% carrot, 15% cabbage").

Similarly, if a survey aggregates food consumed in overly broad categories, these will have to be decomposed and replaced with their food genus equivalents. The proportion of the various food genuses that make up a food consumption entry is up to the person bring their data. For example, IHS4 just records "rice" consumption; we will replace every instance of "rice" with two entries, one of food genus "white rice" and one of food genus "brown rice".