Business Analytics and Decision Making
Business Analytics and Decision Making
Introduction
Business analysis refers to the process of effecting changes in an organizational operation. This is done through a series of activities that are designed to identify an organizational business needs, and reaching a solution that will bring value to the organization. It includes process development and improvement and as well changes in the organizational culture and management practice. This research paper will explore the relationship between the number of website hits and sales in an attempt to make sales forecast based on the number of hits.
The client is involved in the sale of lottery apps for smart phones. The client has identified a trend that suggests a relationship between the sales and website hits in the previous month. Thus, he wishes to test this hypothesis and be able to use the number of hits in a month to be able to forecast a subsequent month’s sale from the previous one. A three months data for the hits and the monthly sales is availed for analysis. We shall thus carry out regression analysis and use the formula in forecasting. The results will then be compared to the actual sales figure and recommendations on how the data can be used in forecasting will be presented. The data are presented below:
In statistical modeling, regression analysis refers to a process of estimating a relationship between variables. It includes various techniques that are used in predicting and analyzing variables. The relationship is usually analyzed between the dependent variable and one or several independent variables. Regression analysis helps in understanding how a dependent variable varies when an independent variable is changed. It gives the conditional expectation of the dependent variable when the independent variable is varied (Montgomery & Vining, 2012). The estimation of the target function is known as regression function. It is widely used in forecasting dependent variables given a conditional change of the independent variables. We carried out a linear regression analysis and the graphical representation of the variables is as below.
Since the number of hits is the independent variable while the sales figure is the dependent variable, we plotted the figures in the table below.
Figure 1: sales plotted as a function of the number of hits
Y (sales) = 0.478x -57.55
The R2 = 0.597
A close examination of the plotted figures reveals that there were two outliers in the data. The month of December and February had a rather an unusual variable. December had so many hits at 1287 compared to the other months yet the sales were lower than the other months that had a relatively lower number of hits. This has the effect of skewing the data that resulting in a lower coefficient. If for instance we were to ignore the variables in December, the correlation coefficient would increase to 0.896 and thus will result to a more accurate forecasting model.
The above regression indicates a positive correlation but not a very strong one. However, given that the correlation is above 0.5, we can use it to estimate the dependent variables. The variables are thus related and an increase in the number of hits will lead to a corresponding increase in the sales. Using the above equation, we predicted the variables in the months of January, February, March and April.
February sales = 0.478(1164)-57.55 = 499
March sales = 0.478(1159)-57.55 = 497
April Sales = 0.478(1298)-57.55 = 563
We then compared the predicated values with the actual values in the tale below
Report
In order to make the model more accurate, we need to understand other factors that affect the sales rather than the number of hits. For instance, there is an explanation as to why the large number of hits in the month of December but relatively low sales as compared to the other months. The inventory levels, prices or any other error that may have affected the efficiency in translating hits to sales. Once we gain a clear understanding of these other factors, we can modify the modeling equation and possibly make more accurate forecasts. The model assumes that most of the website visitors are looking to buy the products. The website is thus assumed to contain a good sales proposition that will motivate turn potential buyers into real buyers.
I would therefore recommend a thorough marketing research and customer feedback program from the client. This will help the client in identifying the other factors that motivate the buyers to purchase the products other than just a visit to the website. There is also a need for the model to incorporate other factors so that the prediction can be more accurate. Thus, would recommend a linear model to be used in the prediction alongside some modifications in the parameters to ensure that the model is accurate. In case the number of visits in a particular month is not consistent with the model, factors that led to the inconsistency should be noted to facilitate a further improvement of the model.
Linear regression models are ideal for linear aligned data that has a strong correlation coefficient. In addition to the model, the analyst should factor in other contributors that have the ability to modify the data and alter the results.
Reference
Montgomery, D. C., Peck, E. A., & Vining, G. G. (2012). Introduction to linear regression analysis. Hoboken, NJ: Wiley.