Disclaimer: This is a test on how the author will take study notes in the future and testing the lay-out of the website. Please do not use the below information or on this website for any subject matters or for use of examinations. Furthermore, the author is not rendered any service in accounting, taxation or similar professional services.
Regression Analysis
I. Overview
Correlation analysis.
The coefficient of correlation.
Correlation between two variables is represented by plotting their values on a graph to form a scatter diagram.
Coefficient of determination.
Regression analysis.
y=a+bx+e
y= dependent variable.
a= y-axis intercept (fixed cost in cost function).
b= slope of the regression line (variable cost in cost function).
x= independent variable.
e= error term (the residual or disturbance term).
II. Correlation Analysis
It varies between -1 and 1.
The objective is to display a population of items for analysis.
The coefficient of correlation is zero.
New factors not included in the model are causing dependent variables to change.
It may be in result of a large part of random chance.
III. Regression (Least Square) Analysis
Multiple regression has more independent variables.
Ratio analysis and trend analysis.
- Graphic method.
- Simple regression.
- High-and low-point method.
Note: multiple regression should not be used because it has more than one independent variable.
Autocorrelation or serial correlation.
Independent variables are correlated with each other.
The coefficient of determination.
The confidence interval.
Standard error of the estimate.
- The errors are normally distributed, and their mean is zero.
- The variance of the errors is constant.
- The independent variables are not correlated with each other.
- No changes occur in the environment.
Constant variance.
- The parameter estimate of the y-intercept and slope also has a normal distribution.
- The estimate of the slope can be tested using a t-test.
- The probability that the error term is greater than zero is equal to the probability that it is less than zero for any observation.
IV. Separating Fixed and Variable Costs
Regression analysis.
Time series or trend analysis.
Monte Carlo simulation.
Bayesian statistics.
V. Other Forecasting Methods
Sensitivity analysis.
Time series analysis.
Trend, cyclical, seasonal, and irregular.
By taking the weighted average over four time periods.
Cross-sectional regression analysis.