Monthly sales from my job that I use to explore OLS regressi

Monthly sales from my job that I use to explore OLS regression

We have many different hearing aids and I would like to see if each hearing aid sells better a certain time of year.

Identify the independent and dependent variables you would use.

The type of hearing aid and the month is independent variables and the price of the hearing aid is the dependent variable.

Identify any anticipated problems with data collection.

I would do over a time period of around 5-10 years to be able to predict a trend, but an anticipated problem is they we have done a few new product releases within the last 5 years and whenever we do a new product release we see an influx of sales. With those new product releases we\'ve done them at different times throughout the year so I\'m hoping over a longer period of time it\'ll even out.

Do the variables you have chosen meet the requirements for a meaningful regression analysis?

Quantitative variables: Monthly sales and quantity of each hearing aid.

Linear - I would only know that one I ran the test, so it is an assumption at this point.

Outliers: I think the product release months could be the outliers that would be recorded, if any other are found those will also be noted.

QUESTION: PLEASE I NEED YOUR COMMENT ON THIS ANALYSIS

Solution

From the case study, I could understand that monthly sales data are available for different types of hearing aids. The objective of the study is to see if there is any significant impact on sales due to the factors: types of hearing aids and specific time of a year (i.e. month).

Therefore, a OLS regression (multivariate) is suitable in this case considering types of hearing aids and month as independent variables and price of the hearing aid is the dependent variable. OLS regression is appropriate in this case as the dependent variable i.e. price of the hearing aid is continuous variable.

As we know that accuracy of the statistical analysis is closely related to the volume of data used. For sufficiently large data set, the error will be minimized. So, a time period of 5 – 10 years (equivalently 60 – 120 months sales data) is sufficient for the regression analysis as for this longer horizon of time, we will take account presence of outliers, seasonal trends, product life cycle, any other external market factors. Also it is mentioned that there are some new product releases in the past 5 years and the business witnessed an influx of sales due to the new product launch. This is a typical market behavior as many product enjoys benefit of the introductory period. So, to even out this effect, it is advisable to launch the new products at different times of a year (i.e. different months).

Assumptions for the OLS Regression to be valid:

Residuals are defined as, Residual = (Original value of the dependent variable – predicted value of the dependent variable)

Outlier Treatment: For this business case, the new product release month could be an outlier. A boxplot of the monthly sales data can be drawn to determine the outliers.