> **来源：[研报客](https://pc.yanbaoke.cn)** ```markdown # Summary of "Processing Food Consumption Data from Household Consumption and Expenditure Surveys (HCES)" ## Core Content This document provides **guidelines for processing food consumption data** collected through **Household Consumption and Expenditure Surveys (HCES)**, in line with the **United Nations Statistical Commission-endorsed IAEG-AG 2018 guidelines**. It is designed to support **national statistical offices (NSOs)** and other organizations in preparing consistent, reliable, and high-quality food data for **poverty, food security, and nutrition analysis**. The guidelines outline a **standardized, step-by-step process** to transform raw food data into usable forms, including **quantities in grams**, **monetary values**, and **dietary energy (kcal)**. They emphasize the importance of **data cleaning, consistency, and transparency** to ensure that food data can be used effectively for macroeconomic and socio-economic analyses. ## Main Views ### 1. **Purpose of the Guidelines** - To standardize the processing of food data from HCES across countries. - To ensure data is consistent, transparent, and ready for use in poverty, food security, and nutrition analysis. - To improve the quality and reliability of statistics derived from food consumption data. ### 2. **Scope of the Guidelines** - Focus on **food consumption modules** within HCES, which include: - In-house consumption - Food away from home (FAFH) - Cover **data cleaning, imputation, conversion, and aggregation**. - Include recommendations on **nutrient conversion tables (NCTs)** and **data documentation**. ### 3. **Key Steps in the Process** - **Step 1**: Gathering input and auxiliary data (e.g., food composition tables, price data, non-food data). - **Step 2**: Data cleaning (checking for negative values, duplicates, consistency, and validity). - **Step 3**: Adjusting and merging data files. - **Step 4**: Cleaning at the food item and unit level (detecting and correcting outliers). - **Step 5**: Imputing monetary values for missing data. - **Step 6**: Converting food quantities into grams. - **Step 7**: Editing after conversion to grams. - **Step 8**: Calculating dietary energy and macronutrients. - **Step 9**: Imputing dietary energy for undefined or missing food items. - **Step 10**: Aggregating and macro-editing data. - **Step 11**: Preparing and sharing the final dataset. ### 4. **Data Processing Principles** - **Transparency and replicability** are essential; each step and decision must be well-documented. - **Consistency** across surveys is encouraged to facilitate time-series and cross-country comparisons. - **Standardization** of data processing methods ensures harmonized results for similar indicators. - **Use of NCTs** is recommended to convert food quantities into nutrient values for dietary energy calculations. ## Key Information ### 1. **Data Collection Modules** - **In-house consumption**: Captures food consumed at home, including quantity and source. - **Food away from home (FAFH)**: Includes food purchased or consumed outside the home, such as meals at restaurants or street vendors. ### 2. **Data Cleaning** - Involves **four stages**: initial checks, consistency checks, outlier detection, and correction. - **Outlier identification** is done using statistical methods such as **box plots** and **median absolute deviation (MAD)**. - **Imputation** is used to replace missing or incorrect data, particularly for **monetary values** and **quantities**. ### 3. **Conversion to Grams** - All food quantities must be converted into **grams**, whether collected in **standard units**, **volume units**, or **non-standard units (NSUs)**. - A **conversion factor library** is used to assist in this process. - **Quality checks** are performed to ensure the accuracy of weight data in grams. ### 4. **Nutrient Conversion Tables (NCTs)** - NCTs are built using **food composition tables (FCTs)** and **databases (FCDBs)**. - They allow for the **conversion of food quantities into nutrient values** (e.g., kilocalories, macronutrients). - NCTs are essential for **dietary energy calculations** and **nutritional analysis**. ### 5. **Data Documentation** - The **Data Documentation Initiative (DDI)** is recommended for documenting the processing steps. - All **decisions and adjustments** made during data cleaning and processing must be recorded for transparency and reproducibility. ### 6. **Implementation and Collaboration** - The guidelines have been **tested and implemented** by organizations such as **SPC, WB, FAO, and COMESA**. - They have been used successfully in **Pacific Island countries** (e.g., Kiribati, Vanuatu, Marshall Islands, Tonga, Palau, Tuvalu, Samoa) in **Household Income and Expenditure Surveys (HIES)**. - The **UN-CEAG** (formerly IAEG-AG) and **international experts** have contributed to the development of these guidelines. ## Structure of the Document ### 1. **Abbreviations** - A list of key acronyms used in the document is provided for clarity. ### 2. **Boxes** - **Box 1**: Guidelines on food data collection (IAEG-AG 2018). - **Box 2**: Practices for an efficient process (joint processing, full processing, consistency, and NCT generation). - Additional boxes provide examples and explanations on specific aspects of data processing, such as outlier detection, unit conversion, and data documentation. ### 3. **Figures** - Illustrate the **flow of food data processing**, **survey modules**, and **data conversion steps**. ### 4. **Tables** - Provide examples of **data entry issues**, **outlier correction decision matrix**, **final data file structure**, and **NCT examples**. ## Conclusion These guidelines are intended to **support NSOs and other organizations** in preparing **high-quality, standardized food data** from HCES. They ensure **efficient and cost-effective data processing**, **consistency in results**, and **improved statistical analysis** for food security and nutrition studies. The process is **modular, adaptable, and transparent**, allowing for **flexibility in different survey designs** while maintaining a **common framework** for data preparation. ```