Multi-Model Regression Analysis Project

00:00 10 marks

You are required to conduct a regression analysis project using multiple models on a real-world dataset. Your project should focus on evaluating the fit of different regression models: linear, polynomial, exponential, and sine models.

Part a

8 minutes 4 marks

Question

Gather a real-world dataset that you believe is suitable for regression analysis using linear, polynomial, exponential, and sine models. Provide a detailed justification for your dataset choice, explaining why it is appropriate for testing these four types of regression models. Include a description of the variables and the source of your data.

Hint Hint

Hint

When justifying your dataset, consider whether it exhibits trends that could be linear, polynomial, or exponential, and if there are cyclical patterns that a sine model might capture. Discuss the nature of the variables and the source’s reliability.

Answer Answer
Mark Scheme Mark Scheme
Explanation Explanation
Formula Booklet

Answer Answer

Complete Answer

A suitable dataset could be monthly global average temperatures from a reputable climate data source (e.g., NASA GISS). This dataset is appropriate because it exhibits both long-term trends (potentially linear or exponential due to climate change) and seasonal cycles (sine wave pattern). Variables are time (months/years) and average temperature.

Mark Scheme Mark Scheme

Answer Explanation Answer Explanation

Formula Booklet

Part b

12 minutes 6 marks

Question

Using your chosen dataset from Part (a), apply linear, polynomial (degree 2), exponential, and sine regression models. For each model, calculate the Sum of Squared Errors (SSE). Compare the SSE values to determine which model provides the best fit for your data. Explain your conclusion, discussing why the model with the lowest SSE is considered the best fit in this context.

Hint Hint

Hint

Utilize technology to perform the regressions and calculate SSE for each model. Remember that SSE is calculated as the sum of the squares of the residuals (the differences between observed and predicted values). A lower SSE generally indicates a better model fit.

Answer Answer
Mark Scheme Mark Scheme
Explanation Explanation
Formula Booklet

Answer Answer

Complete Answer

The best model will be the one with the lowest SSE. For example, if after performing the regressions and calculating SSE, the polynomial model has the lowest SSE, then the polynomial model is the best fit for this dataset based on SSE. The specific SSE values and best model will depend on the dataset used.

Mark Scheme Mark Scheme

Answer Explanation Answer Explanation

Formula Booklet