Problem 2: A government researcher is analyzing the relationship between retail sales and the gross national product (GNP).He also wonders whether there are significant differences in retail sales related to the quarters of the year, compared to the baseline of Q4. He collects ten years of quarterly data.

Year	Quarter	Retail sales (in millions)	GNP (in billions)
2002	Q1	696048	9740.5
2002	Q2	753211	9983.5
2002	Q3	746875	10048.0
2002	Q4	792622	10184.9
2003	Q1	704757	10206.2
2003	Q2	779011	10350.9
2003	Q3	756128	10332.2
2003	Q4	827829	10463.1
2004	Q1	717302	10549.7
2004	Q2	790486	10634.7
2004	Q3	792657	10749.1
2004	Q4	833877	10832.2
2005	Q1	741233	10940.2
2005	Q2	819940	11073.6
2005	Q3	831222	11321.2
2005	Q4	875437	11508.3
2006	Q1	795916	11707.8
2006	Q2	871970	11864.2
2006	Q3	873695	12047.3
2006	Q4	938213	12216.6
2007	Q1	836952	12486.3
2007	Q2	932713	12613.0
2007	Q3	940880	12848.7
2007	Q4	987085	12994.1
2008	Q1	897180	13264.0
2008	Q2	987406	13423.3
2008	Q3	978211	13514.8
2008	Q4	1018775	13683.2
2009	Q1	923997	13859.8
2009	Q2	1016136	14087.6
2009	Q3	1002312	14302.9
2009	Q4	1062803	14489.9
2010	Q1	953358	14520.7
2010	Q2	1032919	14647.3
2010	Q3	1006551	14689.2
2010	Q4	966329	14317.2
2011	Q1	839625	14172.2
2011	Q2	919646	14164.2
2011	Q3	926265	14281.9
2011	Q4	985649	14442.8

1. Load this data however you wish into R.
2. Which variable is the response ?
3. Since the researcher is interested in whether or not the Quarter of the year has an impact on retail sales, write R commands to create the dummy variables you need to recode the "Quarter" column, using Q4 as baseline.
4. Using R, find the best MLR model for the data (disregard Year variable). Is the overall model significant? Justify with ANOVA hypotheses and p-value. Are the individual variables significant? If not, remove the insignificant variables stepwise, until all remaining variables are significant
5. Predict retail sales in quarter 1 if GNP equals $13,000,000,000,000 (13,000 billion)
6. Interpret the meaning of all significant slope coefficients in the "reduced" model.
7. Look at the residual plots. Are there any patterns? Should we worry about the assumption of MLR?