USING HOUSEHOLD CHARACTERISTICS TO PREDICT RESPIRABLE PARTICULATE LEVELS IN RURAL HOUSEHOLDS OF ANDHRA PRADESH, INDIA

摘要

Using Andhra Pradesh, India as a case study, CART and logistic regression were used to assess how household and fuel characteristics can be used to predict whether households have high or low concentrations of respirable particulate matter in kitchen and living areas. Results of both techniques indicate that high concentrations can be classified more accurately than low concentrations. Fuel type is the most important prediction parameter. Indoor kitchens are more likely to have high kitchen concentrations. For living-area concentrations, the specific type of kitchen is less important than whether the kitchen is separate from the living area. Improvements in kitchen ventilation do not occur at the expense of air quality in the living area. Using both kitchen type and kitchen ventilation does not improve prediction over using one of these parameters and a site visit can help inform which to use in future studies.