shell
pip install dpark
python
import dpark
python
ctx = dpark.DparkContext()
python
data = ctx.textFile("data.txt")
python
mapped_data = data.map(lambda x: x.split(","))
filtered_data = mapped_data.filter(lambda x: int(x[2]) > 18)
total_age = filtered_data.map(lambda x: int(x[2])).reduce(lambda x, y: x + y)
grouped_data = mapped_data.groupByKey()
python
result = total_age
print("Total Age:", result)
python
ctx = dpark.DparkContext(master="local[4]", queue="mypriority")
