BUSI650 Predictive model – Multiple Linear Regression Model (MLR)

1. Find datasets from WEB and / or files from MOODLE create your business case/

scenario for MLR 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 input because 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.

6. Create a main chapter for Multiple Linear Regression Model keep these sub headers in your assignment.

a. Business case/scenario etc.

b. Technical details of the model – introduction

c. Data understanding and Data preparation for the model

d. Model development

e. Model comparison on training dataset – Evaluation metrics

f. Final model selection – all assumption checks- Evaluation metrics- SHOW ASSUMPTION CHECKS IN FINAL MODEL.

h. The performance of the final model in Test data set, comparison of real values and test split values (visual). Create a scatter graph between the predictions and the real values.

i. Understanding the final model coefficients- business scenario meaning of the model, in FINDINGS or DISCUSSIONS. Use extra features to analyze the data like gender distribution of dependent variable or scatter of dependent with continuous explanatory variables etc.

j. If you are using categorical variables as explanatory variables (inputs of the model) do not forget dummy encoding (one hot encoding)

k. Conclusion of the model part

4. Max 20 pages, 12 font sizes, consistent style, Fig, Table names, give reference to the Table or figure in discussions before or after the table,1 line spacing only.

5. Adjust figure and table size, be consistent. (Not too big not too small)

6. Adjust the spaces in the page, no redundant inconsistent space

7. Deliver the files via MOODLE, check file size and file format before uploading the file, speak your classmates before submitting, ask for a second review for the final version of the file. File will be graded, but you must submit the excel too, because I need to understand the originality of the work. If you still have a problem, deliver via email, and explain the problem. If needed decrease the file size of excel for submission.