Quantitative Analysis - Dupree
4201 words 17 pagesExecutive Summary
Dupree Fuels Company sells heating oil to residential customers. The company wants to guarantee to its customers that they will not run out of heating oil at any time during the winter months. Factors such as the energy efficiency of homes and the temperatures during winter months have been shown to be important factors related to the amount of heating oil that customers use. The company collected data from a sample of 40 residential customers regarding four variables, oil usage, degree days, a measure of the difference in temperature from 68 degrees Fahrenheit over the period since the last tank fill up, the number of people living in the home, and a home index that measures the energy efficiency of the homes. The …show more content…
Data Characteristics Table 1 shows the descriptive statistics for the data collected from the sample of residential customers. The mean oil usage was 218.05 gallons, while the number of degree days between tank refills was 633.38. In addition, the mean home index for the customers in the sample was 2.75, which indicates that the homes of the customers had a high level of energy efficiency, and the mean number of people in each of the homes was 4.35.
Table 1: Descriptive Statistics | Mean | Standard Deviation | Minimum | Maximum | Oil Usage | 218.05 | 176.70 | 7 | 679 | Degree Days | 633.38 | 381.38 | 54 | 1464 | Home Index | 2.75 | 1.43 | 1 | 5 | Number People | 4.35 | 1.31 | 1 | 7 |
One of the important assumptions in using data to predict some type of outcome, in this case, the prediction of the amount of heating oil used by customers of Dupree Fuels Company, is a lack of multicollinearity. When the variables that are used to predict an outcome have a high level of multicollinearity, which means that they are highly correlated with each other, it is difficult to make predictions with a high level of accuracy because as one variable changes, another variable with a high level of correlation will change in the same way. Table 2 shows the correlation coefficients for the three predictor variables included in the dataset. The correlation coefficients between the three predicator variables are very low. Multicollinearity does not exist between the