W4121: Non-graded Quiz, Wu’s Lectures
Click Here For Solutions
Lec2: Data Modeling
The point of data modeling is to think about what characteristics of the data you will need for your application.
Q1: Data Independence
- What is data independence and why does it matter?
- What are the main two reasons why hierarchical data model is a bad idea?
- Give an example that illustrates the bad idea.
- When could it make sense to use a hierarchical data model?
- What did the graph model try to solve? What limitations did it still have?
Suppose our data adheres to the following schema
animals(id, species, name, gender, keeper_id)
keepers(id, name, birthday)
feedings(keeper_id, animal_id, date, time, food, pounds)
Describe in english what each of the following operations will do:
- filter(feedings, time < 12pm)
- filter(feedings, time < 12pm and pounds > 10)
- project(animals, [name])
- project(animals, [species])
- join(feedings, keepers, keeper_id = id)
- join(animals, feedings, keeper_id = id)
- groupby(feedings, [food], SUM(pounds), COUNT(pounds))
What are the operator symbols (greek symbols) for the above operations?
Describe in english what the following operations do:
- Πfood σtime=12pm(feedings)
- Πname (σtime=12pm(feedings) ⋈id=animal_id σspecies=gorilla(animals))
We have seen that having a schema is helpful because then data can be easily queried. Before the solution described in the paper, Twitter’s event messages DID adhere to a schema of
- Even though the events had a schema, why did data scientists still have trouble analyzing logging data?
- What was the source of this trouble?
- Why do you think Twitter let this happen? Couldn’t they have prevented it from the very beginning?
- What was the main solution to address the above troubles?
Lec 3/4: Query Execution