Advanced Multiple Regression Analysis

Fytokem Products Inc.

Advanced Multiple Regression Analysis

Presentation By:Kamalika SomeKruthik KulkarniRitesh PrasadPankaj Kumar

Case Study

• Canada based company producing pharmaceutical ingredients.• Facing poor sales with domestic customers due to

lack of demand.• Introduction of Tyrostat in the international market

– Success.• Increase in sales by an average of 22%

1) Predicting the Size of Purchase

1) Predicting the Size of Purchase : Scatter Plots

1) Predicting the Size of Purchase

1) Adjusted R-squared is 70%.

2) Company Size is a significant variable.

3) P-value of Cost of delivery and Similar products >0.05, which indicates non-significance of these variables in the model.

Predicting Size of Purchase with Company

Size1) Adjusted R-square is 66%.2) P-value for company size

is <0.05 which indicates significance.

3) Size of Purchase = 23.904 + 1.782 * Company Size

Residual Plot: The most relevant variable

alone Company Size

2) Analysing the response variable - Sales

2) Analysing the response variable – Sales: Scatter Plots

2) Analysing the response variable - Sales

1) Adjusted R-squared is very low.

2) P-value for explanatory variables are >0.05.

3) Exploratory variables do not explain the response variable.

Effect of the variable - Hours worked per Week

Effect of the variable – Number of Customers

3) Measuring the impact of the number of Employees

Sales vs Number of Employees

Tukey’s 4 Quadrant Approach

Sales^2.5 vs (log(Number of Employees)+Number of Employees)

3) Measuring the impact of the number of Employees

1) Adjusted R-squared is 80%.

2) Transformed exploratory variable, log(Number of employees)+Number of employees explains 80% of the variability of response variable.

3) 2.5*Sales=-352961.7 + 86210.2 * log(Number of employees) -477 * Number of employees

Residual vs Fitted

Thank you