Suppose I have a dataset, in which I have questions attempted into 4 categorues. 1…
- User
Suppose I have a dataset, in which I have questions attempted into 4 categorues. 1 confident, in which I felt 3 option could be eliminated, 2 confuse 2 option can be eliminated, 3 bad guess, 1 option can be eliminated. 4 I did not attempt the question. The answer still can be from any 4 option without bias of elimination or not. Now suppose I have 100 questions in a test, and there are say 10 such test. How should I do the statistical analysis or regression analysis such that I could get the number of questions I should Attempt in a test,or percentage of questions I should attempt in a test such that my accuracy is also good and my score is also maximum. 1 question is eaul to 1 score. If a question is marked wrong then negative 0.3 is there. The data give is questions attempted from all 4 categorues. And for each category we have count pf number of cirrected as well as incordect questions answers.
- Assistant
To determine the optimal number or percentage of questions to attempt in order to maximize your expected score while maintaining good accuracy, you can perform the following statistical analysis using your dataset:
1. Calculate Empirical Probabilities for Each Category:
For each category (Confident, Confused, Bad Guess), compute the empirical probability of getting a question correct based on your past performance.
- For each category ( c ): [ p_c = \frac{\text{Number of correct answers in category } c}{\text{Number of questions attempted in category } c} ]
2. Compute Expected Score per Question in Each Category:
Using the scoring system (+1 for a correct answer, -0.3 for an incorrect answer, 0 for unattempted), calculate the expected score per question for each category.
- Expected Score per Question in Category ( c ): [ E[\text{Score per question in category } c] = (1) \times p_c + (-0.3) \times (1 - p_c) = 1.3 p_c - 0.3 ]
3. Determine Which Categories Have a Positive Expected Score:
- If ( E[\text{Score per question in category } c] > 0 ), then attempting questions in that category increases your expected total score.
- If ( E[\text{Score per question in category } c] \leq 0 ), then it's better not to attempt questions in that category, as it would decrease your expected total score.
4. Decide Which Questions to Attempt:
- Attempt all questions in categories where the expected score per question is positive.
- Do not attempt questions in categories where the expected score per question is zero or negative.
5. Estimate the Number or Percentage of Questions to Attempt:
Based on your typical distribution of questions into categories during a test, estimate how many questions you will classify into categories with a positive expected score.
- Example:
- Suppose in a typical test you classify:
- 20 questions as Confident
- 30 questions as Confused
- 20 questions as Bad Guess
- 30 questions as Not Attempted
- If the expected scores per question are:
- Confident: ( E[\text{Score}] > 0 )
- Confused: ( E[\text{Score}] > 0 )
- Bad Guess: ( E[\text{Score}] \leq 0 )
- You should attempt the 50 questions in the Confident and Confused categories.
- Suppose in a typical test you classify:
6. Maximize Expected Total Score:
By focusing on attempting questions only in categories where you have a positive expected score per question, you're maximizing your expected total score.
7. Balance Accuracy and Number of Attempts:
This approach ensures that you're not only maximizing your score but also maintaining good accuracy since you're only attempting questions where you have a statistically significant chance of scoring positively.
Summary:
Performing this analysis allows you to:
- Understand your strengths in different categories based on empirical data.
- Strategically choose which questions to attempt to maximize your expected score.
- Estimate the optimal number or percentage of questions to attempt in future tests.
Note: This strategy assumes that future tests will have a similar distribution of questions and that your performance in each category remains consistent.
By following this statistical approach, you'll be able to make informed decisions on how many questions to attempt in order to achieve the best possible score given your historical performance data.