Linear-Regression

Data fitting using linear regression

1. Building a model for predicting average lifespan using linear regression.

• Requirement 1a: Utilize all 10 features provided in the assignment

  • Train only once for all 10 features on the entire training dataset (train.csv)
  • Express the formula for the linear regression model (predicting $y$ based on 10 features in $X$)
  • Report one result on the test set (test.csv) for the trained model

• Requirement 1b: Construct a model using only 1 feature, find the best-performing model for the optimal outcome

  • Test on the entire set of provided 10 features
  • Requirement: Utilize the 5-fold Cross Validation method to identify the best-performing features.
  • Report 10 corresponding results for 10 models from 5-fold Cross Validation (take the average).

  NO.	Model with 1 feature	RMSE
  1	Adult Mortality
  2	BMI
  	...
  10	Schooling

  • Express the formula for the linear regression model with the best-performing feature (predicting $y$ based on the identified best feature)
    • Report one result on the test set (test.csv) for the best-performing model found

• Requirement 1c: Construct a model, seeking the best-performing model for optimal results

  • Construct m different models (at least 3), each distinct from the models in 1a and 1b
    • Models can be a combination of 2 or more features
    • Models may use standardized or transformed features (squared, cubed, etc.)
    • Models can incorporate features generated from 2 or more different features (addition of 2 features, multiplication of 2 features, etc.)
    • ...
  • Requirement: Utilize the 5-fold Cross Validation method to identify the best-performing model
  • Report m corresponding results m models based on 5-fold Cross Validation (averaged)

  NO.	Models	RMSE
  1	Using 2 features (a, b)
  ...	...
  m	...

  • Express the formula for the best-performing regression model that the student identified
  • Report one result on the test set (test.csv) for the best-performing model found.

• NOTE:

  • The test.csv dataset is only to be used three times, as mentioned earlier.
  • The reported result is a measurement RMSE

2. Report on the results, evaluate, and provide comments on the constructed models (3 points)
   • The report should include personal information (full name, student ID), page count, and references (specifying the documents used for each part of the assignment)
   • List ALL libraries used and the reasons for their utilization
   • Enumerate ALL functions used and describe each of them
   • Report and provide feedback on the results from ALL constructed models (with $1 + (10 + 1) + (m + 1)$ results)
   • With requirements 1b and 1c: explain or state hypotheses for the model that achieved the best result.
   • With requirement 1c: explain the reasons for selecting/designing features for the m models.