Aim: Maximum views on my answers
Definitions
question infancy period: 2 days
developing question: A question that has little to no answers, likely to be within infancy period
developed question: A question that has many answers, likely to be over infancy period
top answers: Top (max 6, scrape limit) answers to a given question
Note
- Code based evidence requires inspection of the code
Experiment 1
Goal: Determine relationship between view count of question and its answers
Results: High question view count leads to that of answers
Part 1
Hypothesis: For developed questions, high views on question results in high views on top answers
Approach:
1. Grab top answers from each of 250 developed questions
2. Obtain datasets for `mean answer view count` and `question's view count`
3. Normalise datasets
4. Use Pearson's r to determine correlation between two datasets
Conclusion: Medium association (0.3<r=0.488<0.5) between question and top answers view count
Evidence: from quora.ex1.scripts import P1; P1().calculate_r()
Part 2
Hypothesis: Pearson's r remains same or improves for developing questions
Approach:
1. Grab top answers from each of 250 developing questions
2. Obtain datasets for `mean answer view count` and `question's view count`
3. Normalise datasets
4. Use Pearson's r to determine correlation between two datasets
5. Compare Pearson's r with that of the developed questions
Conclusion: Pearson's r is higher (r=0.682) for developing questions
Evidence: from quora.ex1.scripts import P2; P2().calculate_r()
Experiment 2
Goal: Determine optimal ranking algorithm verified by simulation on past data.
Results: