Pseudocode to Calculate average using MapReduce

Question

Hi I want to write a MapReduce algorithm in pseudo code to solve the following problem: Given input records in the following format:
address, zip, city, house_value, please calculate the average house value for each zip code. I would really appreciate if you could help me with this..

Answer 1

The easiest would be to use Apache Pig, here is an example of finding an average:

inpt = load 'data.txt' as (address:chararray, zip:chararray, city:chararray, house_value:long);
grp = group inpt by zip;
average = foreach grp generate FLATTEN(group) as (zip), AVG(inpt.house_value) as average_price;
dump average;

For Pseudo Map Reduce code you would need one MAPPER, COMBINER and a REDUCER

MAPPER(record):
    zip_code_key = record['zip'];
    value = {1, record['house_value']};
    emit(zip_code_key, value);

COMBINER(zip_code_key, value_list):
    record_num = 0;
    value_sum = 0;
    foreach (value : value_list) {
       record_num += value[0];
       value_sum += value[1];
    }
    value_out = {record_num, value_sum};
    emit(zip_code_key, value_out);

REDUCER(zip_code_key, value_list):
    record_num = 0;
    value_sum = 0;
    foreach (value : value_list) {
       record_num += value[0];
       value_sum += value[1];
    }
    avg = value_sum / record_num;
    emit(zip_code_key, avg);