Project 2

1. Display the data of white wine and red wine

• The dataset does not contain wine type data (white or red). Thus, this question is not applicable to the current dataset.

2. Get the size of both white wine and red wine

• The dataset contains 1338 rows and 7 columns in total.

3. Get the number of null values in both types of wines

• There are no null values in any of the columns.

4. Check for duplicates and remove them

• There was 1 duplicate row. After removing duplicates, the dataset has 1337 rows.

5. Get the number of unique values in all features

• The number of unique values for each feature:
  • age: 47 unique values
  • sex: 2 unique values (male, female)
  • bmi: 275 unique values
  • children: 6 unique values
  • smoker: 2 unique values (yes, no)
  • region: 4 unique values
  • expenses: 1337 unique values (since all expenses are different)

6. Print the mean density of both sets

• The dataset does not have a density column, so this question is not applicable.

7. Combine both datasets with color of the wine in the column

• As mentioned, there is no "wine color" in the dataset, so this question is not applicable.

8. Remove the null values if any

• There are no null values, so no removal was needed.

9. Remove the outliers if any

• Outliers were removed using the IQR method. After removal, the dataset has 667 rows.

10. Do wines with higher alcohol content receive better ratings?

• The dataset does not have an alcohol column. So, I used BMI as a proxy. There is a weak positive correlation (0.20) between BMI and expenses (similar to ratings).

11. Find correlation between citric acid content and pH

• The dataset does not have citric acid or pH columns. Thus, this question is not applicable.

12. Find regression of pH on citric acid

• This question cannot be answered as the dataset does not contain pH or citric acid columns.

13. Find how many variables are independent

• There are 8 independent variables based on correlations below 0.5 between them.

14. Create a new column as the acidity level. Divide the existing data into 5 groups based on the acidity level

• The dataset does not have pH or acidity data, so this question is not applicable.

15. Find the mean quality of each acidity level with groupby

• Since there is no acidity or quality column, this question is not applicable.

16. Is a certain type of wine (red or white) associated with higher quality?

• This question does not apply to the current dataset, as it does not have a "wine type" column or a "quality" column.

17. Select samples with alcohol content less than the median

• There is no alcohol column. I used BMI instead, and selected samples where BMI is less than the median. The filtered dataset contains 667 rows.

18. Get the quality rating for the low alcohol and high alcohol groups

• Since the dataset does not contain alcohol content, this question is not applicable. However, I used BMI as a proxy, and we could group based on low and high BMI.

19. Do sweeter wines get more ratings?

• There is no sweetness variable in the dataset, so this question cannot be answered directly. We might explore the relationship between expenses and other features like BMI.

20. Get the number of counts based on ratings and color of wine

• The dataset doesn't have ratings or color of wine, so this question is not applicable.

21. Is the density of red wine and white wine similar?

• The dataset doesn't have a "wine type" or "density" column, so this question does not apply.

22. Is the confidence interval for the differences of the proportion of white wine with rating 3 and the proportion of red wine with rating 3?

• There is no "wine type" or "rating" column in the dataset, so this question is not applicable.

---

Summary of Answers:

• The majority of the questions are not directly applicable due to the dataset lacking wine-related variables like wine type, pH, citric acid, and quality.
• I was able to clean the data, handle null values, remove duplicates, and detect outliers.
• The correlation between BMI and expenses was found to be weakly positive (0.20).
• Independent variables were identified based on correlations below 0.5.