LIR 832:A Computer Problem Set on Regression:This problem set is intended to familiarize you with regression in MINITAB and providean initial opportunity to interpret regression output. You will need to download the companysales data set and the accompanying documentation from the web site.1. Load the company sales data set and obtain descriptive statistics for all of the variables inthe data set. What is the definition of each variable? What are the units for eachvariable.2. Lets investigate the factors which influence profits. a. To start, regress market valuation (MV) on sales (MV is the dependent variable,sales is the explanatory variable). Under results get regression equation, table ofcontents, etc. Based on these estimates, what is the relationship between salesand MV (tell me exactly how much MV rise for a $1.00 increase in sales). Testthe hypothesis that MV increases with sales. Can the null be rejected? Now testthe hypothesis that MV rises dollar for dollar with sales (the coefficient would be$1.00). Is there support for this hypothesis? How well does the equation fit thedata (what is r2 and r-bar-square)?Under the regression command, select the fitted line command. Make MVthe Y variable and sales the X variable. What does the resulting graphdepict? What does it tell you about the regression? As a note, you cancopy the graph by right clicking the graph, making it easy to place thegraphs in your homework. Examining the graph, do you see any datawhich might have a large effect on the estimates? Explain your insight.b. Locate the firm which is the extreme value in the graph. What firm is it? Howunusual is the observation (how largest is the next largest MV figure)? Copy theMV2 column into a blank column, label it MV2 (you can do this by cutting andpasting the worksheet). Now replace the MV2 figure for IBM with a missing datavalue (*). Run the regression again. What is the relationship between sales andMV2 in this model. Do a hypothesis test for this relationship. What do you find?Now compare the estimates from (a) with those from (b). How doesomitting IBM affect the estimated coefficients? How does it affect thegoodness of fit measures? Which equation do you prefer. What are theadvantages and disadvantages of dropping extreme data?c. Return to the original data (with IBM) so that we can build a more completeequation. Add assets into the model. What is the relationship between assets andMV? How does adding assets to the model change the effect of sales on MV? Test the hypothesis that sales and assets are positively related to profits? How dothe measures of goodness of fit change, what is the implication of this? Based onthese estimates, do assets play an important role in explaining MV once weaccounted for sales?d. Now add profits to the model. What is the estimated relationship between profitsand MVs? How has adding profits changed the measure of goodness of fit? Howhas this changed the relationship between sales and MV? Between assets andMV (pay attention to both the coefficient, the standard error and the t-statistics)? Why is profits having such a large effect on MV?3. Now lets make a small mistake in our understanding of the relationship between MV andprofits. Generally, we believe that profits drive MV. Higher profits cause investors tobid up the stock price, raising MV. But suppose we make a mistake and take profits asour dependent variable and put MV among the explanatory variables. What is theregression estimate for this model (keep assets and sales in). a. What is the relationship between each of the explanatory variables and profits? IsMV a good predictor of profits (highly statistically significant, large coefficient).b. We might as well also get the effect of profits on the goodness of fit measure tobetter assess its importance. Estimate the equation with profits as the dependentvariable and sales and assets as explanatory variables. How does the addition ofMV to the explanatory variables affect r-square and r-bar-square?c. Reflecting on these results, why is theory – having a clear concept of the casualrelations among your variables – important to estimation of a regressionequation?4. Finally, lets see if profits vary systematically by industry. We have classifications ofindustries into categories such as hi tech, finance and energy. Lets see if high techindustries earn larger profits. We need to create a 0/1 variable where 1 stands for high tech industries and 0indicates some other industry. This is a two step project (tehre are other ways,but this is relatively straightforward). 1. Step one is to code from text to numeric using Code (under DATA). Code from the Sector variable into a new column (call this HiTech). Thefirst and only line of the code command should be arranged so that the‘original value’ is ‘HiTech’ and the ‘new value’ is, leave the other linesblank. The HiTech variable should now have 1's for HiTech industries,*’s elsewhere. 2. Now do a numeric to numeric code for the HiTech column. Put * intothe original value column, and 0 into the new value column. Presto, youshould now have a variable that takes on values of 0 and 1 only.Now repeat the regression in which market valuation is determined by profits,sales and assets but add the HiTech variable to your model. What does being in ahigh tech industry do for market valuation? Is Hi-Tech statistically significant?How does this change the other