Find datasets from WEB

create a business case/ scenario for MLR and change the headers and numbers slightly. Do not forget the final model performance is mostly dependent on selected dataset.

2. Do not use too many random created numbers in the model as inputbecause their correlation (and your model performance will be low) will be lower with the dependent (target variable of the model) variable.

3. Split the dataset into two parts 70% train - 30% test.

4. Use train dataset to obtain final model formula and final inputs, use stepwise approach (clean one parameter at a time)

5. Use normalized dataset with final inputs to examine variable importance.