Week Date PRICE 12PK PRICE_12PK.Ln PRICE 18PK PRICE_18PK.

Ln PRICE 30PK
1 1/1/2003 19.98 2.99 14.10 2.65 15.19
2 1/8/2003 19.98 2.99 18.65 2.93 15.19
3 1/15/2003 This file19.98
and the free Excel 2.99add-in which
18.65 produced the analysis (RegressIt)
2.93 13.87 can be downloade
4 1/22/2003 19.98 2.99 18.65 2.93 12.83
This sample data file contains 52 weeks of price and sales data for 3 carton sizes of beer at a
5 1/29/2003 and quantity-sold
19.98 2.99have all18.65
variables been converted to 2.93 13.16
a per-case (i.e., per-24-can) basis to al
6 2/5/2003 compared 19.98in charts and 2.99 18.65 2.93
model coefficients.  For example, the value 15.19of $19.98 for the price of
7 2/12/2003 sold for19.98
$9.99 in that week,2.99and the18.65
value of 223.5 for cases of 12-packs
2.93 13.92 sold in that week me
8 2/19/2003 20.10 3.00 18.73 2.93 14.42
The variable transformation tool in RegressIt (whose dialog box is pictured below) has been used to c
9 2/26/2003 logs of the
20.12 3.00For example.
original variables. 18.75CASES_12PK.Ln 2.93 13.83
is the natural log of CASES_12PK. The log
10 3/5/2003 suitable20.13
for use in a linear 3.00
regression model.
18.75 2.93 14.50
11 3/12/2003 The forecasts
20.14 of the log-log3.00 18.75 model were2.93
simple regression 13.87
saved to the data sheet, showing up as col
12 3/19/2003 errors were
20.12computed in column
3.00 O for18.75
comparison against2.93 13.64
those of the original linear model. See the
13 3/26/2003 20.12 3.00 13.87 2.63 14.31
The last sheet in the file contains the original data prior to the creation of the transformed variables a
14 4/2/2003 20.13 3.00 14.27 2.66 13.85
15 4/9/2003 20.14 3.00 18.76 2.93 14.20
16 4/16/2003 20.14 3.00 18.77 2.93 13.64
17 4/23/2003 20.13 3.00 13.87 2.63 14.33
18 4/30/2003 20.13 3.00 14.14 2.65 13.14
19 5/7/2003 20.13 3.00 18.76 2.93 13.81
20 5/14/2003 20.13 3.00 18.72 2.93 15.19
21 5/21/2003 20.13 3.00 18.76 2.93 13.13
22 5/28/2003 19.18 2.95 18.76 2.93 13.63
23 6/4/2003 14.78 2.69 18.74 2.93 15.19
24 6/11/2003 16.04 2.78 18.75 2.93 13.89
25 6/18/2003 20.12 3.00 18.75 2.93 14.28
26 6/25/2003 19.75 2.98 18.75 2.93 15.19
27 7/2/2003 19.65 2.98 18.75 2.93 13.12
28 7/9/2003 19.69 2.98 13.79 2.62 13.78
29 7/16/2003 20.12 3.00 13.49 2.60 15.19
30 7/23/2003 20.12 3.00 14.89 2.70 15.19
31 7/30/2003 20.13 3.00 13.94 2.63 15.19
32 8/6/2003 20.14 3.00 13.67 2.62 15.19
33 8/13/2003 15.14 2.72 14.43 2.67 15.19
34 8/20/2003 14.33 2.66 18.75 2.93 15.19
35 8/27/2003 16.24 2.79 18.22 2.90 13.14
36 9/3/2003 19.93 2.99 14.06 2.64 13.45
37 9/10/2003 21.06 3.05 14.43 2.67 13.00
38 9/17/2003 21.19 3.05 19.48 2.97 13.60
39 9/24/2003 21.23 3.06 15.15 2.72 14.46
40 10/1/2003 20.12 3.00 13.79 2.62 14.94
41 10/8/2003 14.73 2.69 14.31 2.66 15.19
42 10/15/2003 14.57 2.68 19.50 2.97 15.19
43 10/22/2003 15.94 2.77 13.85 2.63 15.19
44 10/29/2003 20.70 3.03 14.23 2.66 13.43
45 11/5/2003 19.57 2.97 19.31 2.96 14.37
46 11/12/2003 19.60 2.98 19.29 2.96 15.19
47 11/19/2003 19.94 2.99 13.76 2.62 15.19
48 11/26/2003 21.28 3.06 13.45 2.60 15.19
49 12/3/2003 14.56 2.68 15.13 2.72 15.19
50 12/10/2003 14.39 2.67 19.43 2.97 15.19
51 12/17/2003 16.81 2.82 13.26 2.58 15.19
52 12/24/2003 19.86 2.988707658617 13.92 2.633326654906 15.19
 PRICE_30PK.Ln CASES 12PK CASES_12PK.Ln CASES 18PK CASES_18PK.Ln
2.72 223.5 5.4 439.0 6.1
2.72 215.0 5.4 98.0 4.6
lysis (RegressIt) can be 2.63
downloaded from https://regressit.com.
227.5 5.4 . 70.0 4.2
2.55 244.5 5.5 52.0 4.0
data for 3 carton sizes of beer at a small chain of supermarkets.  The price
er-case (i.e., per-24-can) 2.58 313.5prices and quantities
basis to allow 5.7 to be64.0
directly 4.2
2.72 279.0 5.6
the value of $19.98 for the price of 12-packs in week 1 means that a 12-pack 72.0 4.3
es of 12-packs sold in that2.63week means
238.0that 447 12-packs5.5 were sold. 47.0 3.9
2.67 315.5 5.8 85.0 4.4
is pictured below) has been used to create additional variables that are the natural
2.63 The logged
e natural log of CASES_12PK. 217.0variables will turn
5.4out to be more
59.0 4.1
2.67 209.5 5.3 63.0 4.1
2.63up as column
d to the data sheet, showing 227.0M here, and the5.4 corresponding57.0real 4.0
2.61 See the216.5
e of the original linear model. formula in cell O2. 5.4 54.0 4.0
2.66 169.0 5.1 404.0 6.0
reation of the transformed variables and saved forecasts.
2.63 178.0 5.2 380.0 5.9
2.65 301.5 5.7 65.0 4.2
2.61 266.5 5.6 40.0 3.7
2.66 182.5 5.2 456.0 6.1
2.58 159.0 5.1 176.0 5.2
2.63 285.5 5.7 61.0 4.1
2.72 360.0 5.9 91.0 4.5
2.57 263.0 5.6 59.0 4.1
2.61 443.5 6.1 83.0 4.4
2.72 1101.5 7.0 41.0 3.7
2.63 814.0 6.7 47.0 3.9
2.66 365.0 5.9 84.0 4.4
2.72 510.0 6.2 85.0 4.4
2.57 580.5 6.4 116.0 4.8
2.62 251.0 5.5 544.0 6.3
2.72 237.0 5.5 890.0 6.8
2.72 302.5 5.7 371.0 5.9
2.72 229.5 5.4 557.0 6.3
2.72 188.5 5.2 775.0 6.7
2.72 795.5 6.7 236.0 5.5
2.72 1556.5 7.4 43.0 3.8
2.58 807.5 6.7 63.0 4.1
2.60 243.0 5.5 469.0 6.2
2.56 201.5 5.3 335.0 5.8
2.61 294.0 5.7 75.0 4.3
2.67 220.5 5.4 461.0 6.1
2.70 255.5 5.5 817.0 6.7
2.72 920.5 6.8 200.0 5.3
2.72 730.0 6.6 32.0 3.5
2.72 262.5 5.6 460.0 6.1
2.60 209.5 5.3 751.0 6.6
2.67 283.0 5.6 70.0 4.2
2.72 262.5 5.6 80.0 4.4
2.72 310.0 5.7 523.0 6.3
2.72 278.5 5.6 741.0 6.6
2.72 741.5 6.6 130.0 4.9
2.72 1316.0 7.2 69.0 4.2
2.72 449.0 6.1 493.0 6.2
2.72 505.0 6.2 814.0 6.7
CASES_18PK.Ln.Loglog.model CASES_18PK.Ln.Loglog.model.Resid
6.088 -0.004
4.213 0.372
4.213 0.036
4.213 -0.262
4.213 -0.054
4.213 0.064
4.213 -0.363
4.184 0.259
4.177 -0.099
4.177 -0.034
4.177 -0.134
4.177 -0.188
6.198 -0.197
6.008 -0.068
4.173 0.001
4.170 -0.481
6.198 -0.076
6.069 -0.899
4.173 -0.063
4.188 0.323
4.173 -0.096
4.173 0.245
4.181 -0.467
4.177 -0.327
4.177 0.254
4.177 0.266
4.177 0.577
6.237 0.062
6.385 0.407
5.723 0.194
6.165 0.158
6.296 0.357
5.933 -0.469
4.177 -0.416
4.369 -0.226
6.107 0.043
5.933 -0.119
3.921 0.397
5.606 0.527
6.237 0.468
5.989 -0.691
3.914 -0.448
6.208 -0.077
6.027 0.595
3.980 0.269
3.987 0.395
6.252 0.008
6.405 0.203
5.615 -0.748
3.938 0.296
6.500 -0.299
6.174 0.528
CASES_18PK.Ln.Loglog.real.errors CASES 30PK CASES_30PK.Ln
-1.584 55.00 4.01 This sample data file contains 52 weeks of pri
30.452 66.75 4.20 carton sizes of beer at a small chain of superm
2.452 242.00 5.49 quantity-sold variables have all been converte
can) basis to allow prices and quantities to be
-15.548 488.50 6.19 and model coefficients.  For example, the valu
-3.548 308.75 5.73 12-packs in week 1 means that a 12-pack sold
4.452 111.75 4.72 the value of 223.5 for cases of 12-packs sold in
-20.548 252.50 5.53 12-packs were sold.
19.363 221.25 5.40 Variable names have not yet been assigned, so the
-6.169 245.25 5.50 done is to highlight the data range (including the t
-2.169 148.50 5.00 names) and hit the Create Variable Name on the R
use the Create From Selection command on the Fo
-8.169 229.75 5.44
-11.169 312.00 5.74 Visit RegressIt.com to see the analysis of the data
-87.951 96.75 4.57
-26.564 123.25 4.81
0.063 200.50 5.30
-24.705 359.75 5.89
-35.951 113.50 4.73
-256.294 136.50 4.92
-3.937 225.50 5.42
25.127 122.25 4.81
-5.937 443.75 6.10
18.063 322.75 5.78
-24.403 53.00 3.97
-18.169 140.75 4.95
18.831 210.75 5.35
19.831 110.50 4.71
50.831 568.25 6.34
32.594 115.50 4.75
297.325 58.75 4.07
65.312 77.25 4.35
81.378 66.25 4.19
232.728 50.00 3.91
-141.276 46.50 3.84
-22.169 65.75 4.19
-15.985 252.75 5.53
19.943 179.00 5.19
-42.276 226.25 5.42
24.555 288.50 5.66
188.812 114.25 4.74
305.594 70.00 4.25
-199.004 47.75 3.87
-18.099 98.75 4.59
-36.734 77.00 4.34
336.711 160.50 5.08
 16.501 143.50 4.97
26.128 133.00 4.89
4.071 68.75 4.23
136.406 81.75 4.40
-144.610 56.25 4.03
17.678 68.75 4.23
-172.111 49.25 3.90
333.777 76.50 4.34
 a file contains 52 weeks of price and sales data for 3
eer at a small chain of supermarkets.  The price and
riables have all been converted to a per-case (i.e., per-24-
ow prices and quantities to be directly compared in charts
fficients.  For example, the value of $19.98 for the price of
k 1 means that a 12-pack sold for $9.99 in that week, and
.5 for cases of 12-packs sold in that week means that 447
old.
ave not yet been assigned, so the first thing that needs to be
ht the data range (including the top row with the variable
he Create Variable Name on the RegressIt menu (or equivalently,
om Selection command on the Formula menu).
m to see the analysis of the data and download the software.
Descriptive Statistics Statistics of sales and prices
Variable # Fitted Mean Median Std.Dev. Root.M.Sqr. Std.Err.Mean Minimum
CASES_12PK 52 399.163 272.500 302.555 499.110 41.957 159.000
CASES_18PK 52 256.673 88.000 258.825 362.744 35.893 32.000
CASES_30PK 52 165.043 122.750 119.625 203.160 16.589 46.500
PRICE_12PK 52 19.088 19.980 2.088 19.199 0.290 14.330
PRICE_18PK 52 16.725 18.650 2.411 16.894 0.334 13.260
PRICE_30PK 52 14.379 14.395 0.806 14.401 0.112 12.830

Autocorrelations
Variable Lag 1 Lag 2 Lag 3 Lag 4
CASES_12PK 0.478 -0.035 -0.071 -0.087
CASES_18PK 0.340 -0.160 0.225 0.475
CASES_30PK 0.237 -0.015 0.040 -0.028
PRICE_12PK 0.544 -0.007 -0.282 -0.254
PRICE_18PK 0.306 -0.141 0.173 0.430
PRICE_30PK 0.341 0.069 0.109 -0.089

Series Plots
.
CASES_12PK (n = 52, mean = 399.163, slope = 6.985)
2000

1500

1000

500

0
0 10 20 30 40
.
CASES_18PK (n = 52, mean = 256.673, slope = 7.286)
1000
800
600
400
200
0
0 10 20 30 40
.
CASES_30PK (n = 52, mean = 165.043, slope = -3.076)
600
500
400
300
200
100
0
0 10 20 30 40
.
PRICE_12PK (n = 52, mean = 19.088, slope = -0.050)
22
21
20
19
18
17
16
15
14
0 10 20 30 40
 20
19
18
17
16
15
14
0 10 20 30 40
.
PRICE_18PK (n = 52, mean = 16.725, slope = -0.060)
20
19
18
17
16
15
14
13
0 10 20 30 40
.
PRICE_30PK (n = 52, mean = 14.379, slope = 0.019)
15.5
15
14.5
14
13.5
13
12.5
0 10 20 30 40

Histogram Plots Editable

.
CASES_12PK (n=52, mean=399.163) CASES_18PK (n
35 35
30 30
25 25
20 20
15 15
10 10
5 5
0 0
Min = 159.000 Midpoint = 857.750 Max = 1,557 Min = 32.000 Midp

.
CASES_30PK (n=52, mean=165.043) PRICE_12PK (n
20 35
30
15 25
20
10
15
5 10
5
0 0
Min = 46.500 Midpoint = 307.375 Max = 568 Min = 14.330 Midp

.
PRICE_18PK (n=52, mean=16.725) PRICE_30PK (n
25 25
20 20
15 15
10 10
5 5
0 0
Min = 13.260 Midpoint = 16.380 Max = 19.5 Min = 12.830 Midp
 10 10
5 5
0 0
Min = 13.260 Midpoint = 16.380 Max = 19.5 Min = 12.830 Midp

Correlation Matrix (n=52)

Variable CASES_12PK
CASES_12PK 1.000 CASES_18PK
CASES_18PK -0.295 1.000 CASES_30PK
CASES_30PK -0.210 -0.458 1.000 PRICE_12PK
PRICE_12PK -0.859 0.255 0.329 1.000 PRICE_18PK
PRICE_18PK 0.241 -0.866 0.521 -0.084 1.000 PRICE_30PK
PRICE_30PK 0.300 0.294 -0.807 -0.364 -0.251 1.000

Scatterplots Editable
.
CASES_12PK vs. CASES_12PK vs.
CASES_12PK CASES_18PK
r = 1.000, r-squared = 1.000 r = -0.295, r-squared = 0.087
1,557 1,557

858 858

159 159
159 858 1,557 32 461 890
.
CASES_18PK vs. CASES_18PK vs.
CASES_12PK CASES_18PK
r = -0.295, r-squared = 0.087 r = 1.000, r-squared = 1.000
890 890

461 461

32 32
159 858 1,557 32 461 890
.
CASES_30PK vs. CASES_30PK vs.
CASES_12PK CASES_18PK
r = -0.210, r-squared = 0.044 r = -0.458, r-squared = 0.210
568 568

307 307

47 47
159 858 1,557 32 461 890
 307 307

47 47
159 858 1,557 32 461 890
.
PRICE_12PK vs. PRICE_12PK vs.
CASES_12PK CASES_18PK
r = -0.859, r-squared = 0.738 r = 0.255, r-squared = 0.065
21.28 21.28

17.81 17.81

14.33 14.33
159 858 1,557 32 461 890
.
PRICE_18PK vs. PRICE_18PK vs.
CASES_12PK CASES_18PK
r = 0.241, r-squared = 0.058 r = -0.866, r-squared = 0.751
19.50 19.50

16.38 16.38

13.26 13.26
159 858 1,557 32 461 890
.
PRICE_30PK vs. PRICE_30PK vs.
CASES_12PK CASES_18PK
r = 0.300, r-squared = 0.090 r = 0.294, r-squared = 0.086
15.19 15.19

14.01 14.01

12.83 12.83
159 858 1,557 32 461 890

End of Output
 Color Font
Maximum
1,557 Here the data analysis procedure has been applied to all of the original sales and price variables, and
890.000 series plots and scatterplots. The series plots show that for each carton size the number of cases so
568.250 where there is a significant price reduction, and the correlations and scatterplots similarly show a st
21.280
between price and sales of the same carton size.
19.500
SCROLL DOWN TO THE BOTTOM OF THIS SHEET TO SEE THE REMAINING OUTPUT: HISTOGRAM PLOT
15.190
AND THE SCATTERPLOT MATRIX.

This analysis and the regression model on the "Linear price-demand model" worksheet were run pri
transformed variables, hence those do not appear in the dialog boxes (yet). A customized analysis n
sales" ) has been entered in place of the default name ("Data Analysis 1"). The name appears in the
date/time/user/analysis stamp at the top of the worksheet and it also serves as the sheet name.

The "Colors" and "Fonts" options have also been turned on for this sheet via the buttons on the Regr
correlations to be color-coded for sign and absolute magnitude and also to have bolder fonts for larg
can be toggled on and off while viewing a model sheet.

= 399.163, slope = 6.985)

30 40 50 60

= 256.673, slope = 7.286)

30 40 50 60

165.043, slope = -3.076)

30 40 50 60

= 19.088, slope = -0.050)

30 40 50 60
 30 40 50 60

= 16.725, slope = -0.060)

30 40 50 60

= 14.379, slope = 0.019)

30 40 50 60

CASES_18PK (n=52, mean=256.673)
5
0
5
0
5
0
5
0
Min = 32.000 Midpoint = 461.000 Max = 890

PRICE_12PK (n=52, mean=19.088)
5
0
5
0
5
0
5
0
Min = 14.330 Midpoint = 17.805 Max = 21.3

PRICE_30PK (n=52, mean=14.379)
5
0
5
0
5
0
Min = 12.830 Midpoint = 14.010 Max = 15.2
 0
5
0
Min = 12.830 Midpoint = 14.010 Max = 15.2

CASES_12PK vs. CASES_12PK vs.

CASES_30PK PRICE_12PK
ared = 0.087 r = -0.210, r-squared = 0.044 r = -0.859, r-squared = 0.738
1,557 1,557

858 858

159 159
61 890 47 307 568 14.33 17.81 21.28

CASES_18PK vs. CASES_18PK vs.

CASES_30PK PRICE_12PK
ared = 1.000 r = -0.458, r-squared = 0.210 r = 0.255, r-squared = 0.065
890 890

461 461

32 32
1 890 47 307 568 14.33 17.81 21.28

CASES_30PK vs. CASES_30PK vs.

CASES_30PK PRICE_12PK
ared = 0.210 r = 1.000, r-squared = 1.000 r = 0.329, r-squared = 0.108
568 568

307 307

47 47
1 890 47 307 568 14.33 17.81 21.28
 307 307

47 47
1 890 47 307 568 14.33 17.81 21.28

PRICE_12PK vs. PRICE_12PK vs.

CASES_30PK PRICE_12PK
ared = 0.065 r = 0.329, r-squared = 0.108 r = 1.000, r-squared = 1.000
21.28 21.28

17.81 17.81

14.33 14.33
61 890 47 307 568 14.33 17.81 21.28

PRICE_18PK vs. PRICE_18PK vs.

CASES_30PK PRICE_12PK
ared = 0.751 r = 0.521, r-squared = 0.272 r = -0.084, r-squared = 0.007
19.50 19.50

16.38 16.38

13.26 13.26
61 890 47 307 568 14.33 17.81 21.28

PRICE_30PK vs. PRICE_30PK vs.

CASES_30PK PRICE_12PK
ared = 0.086 r = -0.807, r-squared = 0.651 r = -0.364, r-squared = 0.132
15.19 15.19

14.01 14.01

12.83 12.83
61 890 47 307 568 14.33 17.81 21.28
 1234 Comment TrendLines Observation # 10.000 6/6/20 2:01 PM + FACDS414 + Beer_sales_

ginal sales and price variables, and the output includes the optional
carton size the number of cases sold increases dramatically in weeks
nd scatterplots similarly show a strong negative relationship

INING OUTPUT: HISTOGRAM PLOTS, THE CORRELATION MATRIX,

nd model" worksheet were run prior to the creation of the

xes (yet). A customized analysis name ("Statistics of prices and
ysis 1"). The name appears in the bitmapped
also serves as the sheet name.

s sheet via the buttons on the RegressIt ribbon. This causes the
d also to have bolder fonts for larger absolute values. This feature
 CASES_12PK vs. CASES_12PK vs.
PRICE_18PK PRICE_30PK
738 r = 0.241, r-squared = 0.058 r = 0.300, r-squared = 0.090
1,557 1,557

858 858

159 159
21.28 13.26 16.38 19.50 12.83 14.01 15.19

CASES_18PK vs. CASES_18PK vs.

PRICE_18PK PRICE_30PK
65 r = -0.866, r-squared = 0.751 r = 0.294, r-squared = 0.086
890 890

461 461

32 32
21.28 13.26 16.38 19.50 12.83 14.01 15.19

CASES_30PK vs. CASES_30PK vs.

PRICE_18PK PRICE_30PK
08 r = 0.521, r-squared = 0.272 r = -0.807, r-squared = 0.651
568 568

307 307

47 47
21.28 13.26 16.38 19.50 12.83 14.01 15.19
 307 307

47 47
21.28 13.26 16.38 19.50 12.83 14.01 15.19

PRICE_12PK vs. PRICE_12PK vs.

PRICE_18PK PRICE_30PK
00 r = -0.084, r-squared = 0.007 r = -0.364, r-squared = 0.132
21.28 21.28

17.81 17.81

14.33 14.33
21.28 13.26 16.38 19.50 12.83 14.01 15.19

PRICE_18PK vs. PRICE_18PK vs.

PRICE_18PK PRICE_30PK
007 r = 1.000, r-squared = 1.000 r = -0.251, r-squared = 0.063
19.50 19.50

16.38 16.38

13.26 13.26
21.28 13.26 16.38 19.50 12.83 14.01 15.19

PRICE_30PK vs. PRICE_30PK vs.

PRICE_18PK PRICE_30PK
132 r = -0.251, r-squared = 0.063 r = 1.000, r-squared = 1.000
15.19 15.19

14.01 14.01

12.83 12.83
21.28 13.26 16.38 19.50 12.83 14.01 15.19
M + FACDS414 + Beer_sales_1.xlsx + Data + RegressItPC 2020.03.04
6/6/20 2:01 PM + FACDS414 + Beer_sales_1.xlsx + Data + RegressItPC 2020.03.04
Descriptive Statistics Statistics of logged variables
Variable # Fitted Mean Median Std.Dev. Root.M.Sqr. Std.Err.Mean Minimum
CASES_12PK.Ln 52 5.802 5.607 0.564 5.829 0.078 5.069
CASES_18PK.Ln 52 5.014 4.477 1.054 5.122 0.146 3.466
CASES_30PK.Ln 52 4.876 4.810 0.680 4.922 0.094 3.839
PRICE_12PK.Ln 52 2.942 2.995 0.119 2.945 0.017 2.662
PRICE_18PK.Ln 52 2.806 2.926 0.148 2.810 0.021 2.585
PRICE_30PK.Ln 52 2.664 2.667 0.057 2.665 0.007856 2.552

Autocorrelations
Variable Lag 1 Lag 2 Lag 3 Lag 4
CASES_12PK.Ln 0.550 0.049 0.008 -0.025
CASES_18PK.Ln 0.367 -0.144 0.225 0.443
CASES_30PK.Ln 0.414 0.119 0.116 -0.027
PRICE_12PK.Ln 0.547 -0.011 -0.272 -0.244
PRICE_18PK.Ln 0.315 -0.141 0.178 0.432
PRICE_30PK.Ln 0.337 0.064 0.100 -0.097

Series Plots
.
CASES_12PK.Ln (n = 52, mean = 5.802, slope = 0.015)
7.5
7
6.5
6
5.5
5
0 10 20 30 40
.
CASES_18PK.Ln (n = 52, mean = 5.014, slope = 0.029)
7
6.5
6
5.5
5
4.5
4
3.5
3
0 10 20 30 40
.
CASES_30PK.Ln (n = 52, mean = 4.876, slope = -0.019)
6.5
6
5.5
5
4.5
4
3.5
0 10 20 30 40
.
PRICE_12PK.Ln (n = 52, mean = 2.942, slope = -0.002940)
3.15
3.05
2.95
2.85
2.75
2.65
0 10 20 30 40
 2.95
2.85
2.75
2.65
0 10 20 30 40
.
PRICE_18PK.Ln (n = 52, mean = 2.806, slope = -0.003743)
3.05
2.95
2.85
2.75
2.65
2.55
0 10 20 30 40
.
PRICE_30PK.Ln (n = 52, mean = 2.664, slope = 0.001323)
2.75

2.7

2.65

2.6

2.55
0 10 20 30 40

Histogram Plots Editable

.
CASES_12PK.Ln (n=52, mean=5.802) CASES_18PK.Ln
16 16
12 12
8 8
4 4
0 0
Min = 5.069 Midpoint = 6.210 Max = 7.35 Min = 3.466 Midp

.
CASES_30PK.Ln (n=52, mean=4.876) PRICE_12PK.Ln
10 35
8 30
25
6 20
4 15
10
2 5
0 0
Min = 3.839 Midpoint = 5.091 Max = 6.34 Min = 2.662 Midp

.
PRICE_18PK.Ln (n=52, mean=2.806) PRICE_30PK.Ln
25 25
20 20
15 15
10 10
5 5
0 0
Min = 2.585 Midpoint = 2.778 Max = 2.97 Min = 2.552 Midp
 10 10
5 5
0 0
Min = 2.585 Midpoint = 2.778 Max = 2.97 Min = 2.552 Midp

Correlation Matrix (n=52)

Variable CASES_12PK.Ln
CASES_12PK.Ln 1.000 CASES_18PK.Ln
CASES_18PK.Ln -0.346 1.000 CASES_30PK.Ln
CASES_30PK.Ln -0.264 -0.520 1.000 PRICE_12PK.Ln
PRICE_12PK.Ln -0.851 0.254 0.436 1.000 PRICE_18PK.Ln
PRICE_18PK.Ln 0.274 -0.942 0.577 -0.082 1.000 PRICE_30PK.Ln
PRICE_30PK.Ln 0.303 0.276 -0.859 -0.363 -0.254 1.000

Scatterplots Editable
.
CASES_12PK.Ln vs. CASES_12PK.Ln vs.
CASES_12PK.Ln CASES_18PK.Ln
r = 1.000, r-squared = 1.000 r = -0.346, r-squared = 0.120
7.35 7.35

6.21 6.21

5.07 5.07
5.07 6.21 7.35 3.47 5.13 6.79
.
CASES_18PK.Ln vs. CASES_18PK.Ln vs.
CASES_12PK.Ln CASES_18PK.Ln
r = -0.346, r-squared = 0.120 r = 1.000, r-squared = 1.000
6.79 6.79

5.13 5.13

3.47 3.47
5.07 6.21 7.35 3.47 5.13 6.79
.
CASES_30PK.Ln vs. CASES_30PK.Ln vs.
CASES_12PK.Ln CASES_18PK.Ln
r = -0.264, r-squared = 0.070 r = -0.520, r-squared = 0.270
6.34 6.34

5.09 5.09

3.84 3.84
5.07 6.21 7.35 3.47 5.13 6.79
 5.09 5.09

3.84 3.84
5.07 6.21 7.35 3.47 5.13 6.79
.
PRICE_12PK.Ln vs. PRICE_12PK.Ln vs.
CASES_12PK.Ln CASES_18PK.Ln
r = -0.851, r-squared = 0.723 r = 0.254, r-squared = 0.064
3.058 3.058

2.860 2.860

2.662 2.662
5.07 6.21 7.35 3.47 5.13 6.79
.
PRICE_18PK.Ln vs. PRICE_18PK.Ln vs.
CASES_12PK.Ln CASES_18PK.Ln
r = 0.274, r-squared = 0.075 r = -0.942, r-squared = 0.888
2.970 2.970

2.778 2.778

2.585 2.585
5.07 6.21 7.35 3.47 5.13 6.79
.
PRICE_30PK.Ln vs. PRICE_30PK.Ln vs.
CASES_12PK.Ln CASES_18PK.Ln
r = 0.303, r-squared = 0.092 r = 0.276, r-squared = 0.076
2.721 2.721

2.636 2.636

2.552 2.552
5.07 6.21 7.35 3.47 5.13 6.79

End of Output
 Color Font
Maximum
7.350
Here are the corresponding statistics and plots for the log
6.791
correlation between log sales and log price is larger in ma
6.343 carton size than it was for the unlogged variables.
3.058
2.970 The unexplained variations seen in the scatterplots of log
2.721 are also more consistent in magnitude for high and low pr
to the bottom of this to sheet to see the scatterplots.)

n = 5.802, slope = 0.015)

30 40 50 60

n = 5.014, slope = 0.029)

30 40 50 60

n = 4.876, slope = -0.019)

30 40 50 60

= 2.942, slope = -0.002940)

30 40 50 60
 30 40 50 60

= 2.806, slope = -0.003743)

30 40 50 60

= 2.664, slope = 0.001323)

30 40 50 60

CASES_18PK.Ln (n=52, mean=5.014)
6
2
8
4
0
Min = 3.466 Midpoint = 5.128 Max = 6.79

PRICE_12PK.Ln (n=52, mean=2.942)
5
0
5
0
5
0
5
0
Min = 2.662 Midpoint = 2.860 Max = 3.06

PRICE_30PK.Ln (n=52, mean=2.664)
5
0
5
0
5
0
Min = 2.552 Midpoint = 2.636 Max = 2.72
 0
5
0
Min = 2.552 Midpoint = 2.636 Max = 2.72

s. CASES_12PK.Ln vs. CASES_12PK.Ln vs.

CASES_30PK.Ln PRICE_12PK.Ln
ared = 0.120 r = -0.264, r-squared = 0.070 r = -0.851, r-squared = 0.723
7.35 7.35

6.21 6.21

5.07 5.07
13 6.79 3.84 5.09 6.34 2.662 2.860 3.058

vs. CASES_18PK.Ln vs. CASES_18PK.Ln vs.

CASES_30PK.Ln PRICE_12PK.Ln
ared = 1.000 r = -0.520, r-squared = 0.270 r = 0.254, r-squared = 0.064
6.79 6.79

5.13 5.13

3.47 3.47
13 6.79 3.84 5.09 6.34 2.662 2.860 3.058

s. CASES_30PK.Ln vs. CASES_30PK.Ln vs.

CASES_30PK.Ln PRICE_12PK.Ln
ared = 0.270 r = 1.000, r-squared = 1.000 r = 0.436, r-squared = 0.190
6.34 6.34

5.09 5.09

3.84 3.84
13 6.79 3.84 5.09 6.34 2.662 2.860 3.058
 5.09 5.09

3.84 3.84
13 6.79 3.84 5.09 6.34 2.662 2.860 3.058

s. PRICE_12PK.Ln vs. PRICE_12PK.Ln vs.

CASES_30PK.Ln PRICE_12PK.Ln
ared = 0.064 r = 0.436, r-squared = 0.190 r = 1.000, r-squared = 1.000
3.058 3.058

2.860 2.860

2.662 2.662
13 6.79 3.84 5.09 6.34 2.662 2.860 3.058

s. PRICE_18PK.Ln vs. PRICE_18PK.Ln vs.

CASES_30PK.Ln PRICE_12PK.Ln
ared = 0.888 r = 0.577, r-squared = 0.333 r = -0.082, r-squared = 0.007
2.970 2.970

2.778 2.778

2.585 2.585
13 6.79 3.84 5.09 6.34 2.662 2.860 3.058

s. PRICE_30PK.Ln vs. PRICE_30PK.Ln vs.

CASES_30PK.Ln PRICE_12PK.Ln
ared = 0.076 r = -0.859, r-squared = 0.738 r = -0.363, r-squared = 0.131
2.721 2.721

2.636 2.636

2.552 2.552
13 6.79 3.84 5.09 6.34 2.662 2.860 3.058
 1234 TrendLines Observation # 10.000 6/6/20 2:19 PM + FACDS414 + Beer_sales_

ding statistics and plots for the logged variables. The

g sales and log price is larger in magnitude for each
or the unlogged variables.

tions seen in the scatterplots of log sales vs. log price

nt in magnitude for high and low prices. (Scroll down
sheet to see the scatterplots.)
 CASES_12PK.Ln vs. CASES_12PK.Ln vs.
PRICE_18PK.Ln PRICE_30PK.Ln
723 r = 0.274, r-squared = 0.075 r = 0.303, r-squared = 0.092
7.35 7.35

6.21 6.21

5.07 5.07
3.058 2.585 2.778 2.970 2.552 2.636 2.721

CASES_18PK.Ln vs. CASES_18PK.Ln vs.

PRICE_18PK.Ln PRICE_30PK.Ln
64 r = -0.942, r-squared = 0.888 r = 0.276, r-squared = 0.076
6.79 6.79

5.13 5.13

3.47 3.47
3.058 2.585 2.778 2.970 2.552 2.636 2.721

CASES_30PK.Ln vs. CASES_30PK.Ln vs.

PRICE_18PK.Ln PRICE_30PK.Ln
90 r = 0.577, r-squared = 0.333 r = -0.859, r-squared = 0.738
6.34 6.34

5.09 5.09

3.84 3.84
3.058 2.585 2.778 2.970 2.552 2.636 2.721
 5.09 5.09

3.84 3.84
3.058 2.585 2.778 2.970 2.552 2.636 2.721

PRICE_12PK.Ln vs. PRICE_12PK.Ln vs.

PRICE_18PK.Ln PRICE_30PK.Ln
00 r = -0.082, r-squared = 0.007 r = -0.363, r-squared = 0.131
3.058 3.058

2.860 2.860

2.662 2.662
3.058 2.585 2.778 2.970 2.552 2.636 2.721

PRICE_18PK.Ln vs. PRICE_18PK.Ln vs.

PRICE_18PK.Ln PRICE_30PK.Ln
007 r = 1.000, r-squared = 1.000 r = -0.254, r-squared = 0.065
2.970 2.970

2.778 2.778

2.585 2.585
3.058 2.585 2.778 2.970 2.552 2.636 2.721

PRICE_30PK.Ln vs. PRICE_30PK.Ln vs.

PRICE_18PK.Ln PRICE_30PK.Ln
131 r = -0.254, r-squared = 0.065 r = 1.000, r-squared = 1.000
2.721 2.721

2.636 2.636

2.552 2.552
3.058 2.585 2.778 2.970 2.552 2.636 2.721
M + FACDS414 + Beer_sales_4.xlsx + Data + RegressItPC 2020.03.04
6/6/20 2:19 PM + FACDS414 + Beer_sales_4.xlsx + Data + RegressItPC 2020.03.04
Descriptive Statistics Statistics of real errors
Variable # Fitted Mean Median Std.Dev. Root.M.Sqr. Std.Err.Mean
CASES_18PK.Ln.Loglog.real.errors 52 18.536 -0.760 117.923 118.246 16.353

End of Output
Here are the summary statistics for the real errors made by the lo
model. The real error in a given week is computed as the actual v
log units), i.e., actual value minus "unlogged" forecast.
The root-mean-square of these real errors is 118.246, which is noticeab
127.994 yielded by the simple linear price-demand model. Thus, the m
not only produces more satisfactory results in log units (in terms of satis
linear regression model), but it also produces more accurate and more l
units.
 Hit Show-All or click +/- signs in sidebar to show or hide output.
Minimum Maximum
-256.294 336.711

the real errors made by the log-log simple regression

ek is computed as the actual value minus EXP(forecast in
unlogged" forecast.
rors is 118.246, which is noticeably less than the value of
rice-demand model. Thus, the model fitted to the logged data
sults in log units (in terms of satisfying the assumptions of a
oduces more accurate and more logical forecasting results in real
s in sidebar to show or hide output. 8.000 6/6/20 2:42 PM + FACDS414
6/6/20 2:42 PM + FACDS414 + Beer_sales_10.xlsx + Data + RegressItPC 2020.03.04
6/6/20 2:42 PM + FACDS414 + Beer_sales_10.xlsx + Data + RegressItPC 2020.03.04
Model: Linear price-demand model
Dependent Variable: CASES_18PK

R-Squared Adj.R-Sqr. Std.Err.Reg. Std.Dep.Var. # Fitted # Missing Critical t

0.751 0.746 130.529 258.825 52 0 2.009

Variable Coefficient Std.Err. t-Statistic P-value Lower95% Upper95% VIF

Constant 1,812 128.070 14.150 0.000 1,555 2,069 0.000
PRICE_18PK -93.007 7.581 -12.269 0.000 -108.234 -77.781 1.000

PRICE_18PK StdErrMean StdErrFcst Predicted Lower 95% Upper 95%

Linear price-demand
13.260 31.898 model134.369
for CASES_18PK (1 variable,309.018
578.908 n=52) 848.797
Predicted CASES_18PK
14.820 23.154 = 1,812 - 93.007*PRICE_18PK
132.566 433.816 167.549 700.083
1000 16.380 18.289 131.804 288.725 23.990 553.460
17.940 20.311 132.099 143.633 -121.696 408.963
800 19.500 27.754 133.447 -1.458 -269.493 266.578
600
400 Upper 95%
200 Predicted
Lower 95%
0
-200
-400
13 14 15 16 17 18 19 20
PRICE_18PK

Mean Error RMSE MAE Minimum Maximum MAPE A-D* stat

Fitted (n=52) 0.000 127.994 82.495 -321.061 332.484 42.8% 2.59 (P=0.000)

Lag 1 2 3 4
Autocorrelation -0.222 0.167 -0.186 0.196
StdErrorsFromZero -1.583 1.180 -1.302 1.357
Durbin-Watson 2.336

Actual and Predicted -vs- Observation #

1.000
Linear price-demand model for CASES_18PK (1 variable, n=52)
1000

800

600

400

200

-200 0 10 20 30 40 50 60
 Residual -vs- Observation #
Linear price-demand model for CASES_18PK (1 variable, n=52)
400
300
200
100
0
-100
-200
-300
-400
Lag 1 autocorrelation = -0.22

Residual -vs- Predicted

Linear price-demand model for CASES_18PK (1 variable, n=52)
400
300
200
100
0
-100
-200
-300
-400
-100 0 100 200 300 400 500 600 700

Histogram of Residuals
Linear price-demand model for CASES_18PK (1 variable, n=52)
18
16
14
12
10
8 Actual
6 Normal
4
2
0
4 0 06 7 2 38 0 4 70 3 6 0 2 6 8 3 4 0 3 4 68 0 2 3 6 7 0 0 4 3 8 7 2 0 6 4 0
-3 -3 -2 -2 -2 -1 -1 -1 - - 1 1 1 2 2 2 3 3

Normality test (A-D*): P < 0.001

 4
2
0
4 0 06 7 2 38 0 4 70 3 6 0 2 6 8 3 4 0 3 4 68 0 2 3 6 7 0 0 4 3 8 7 2 0 6 4 0
-3 -3 -2 -2 -2 -1 -1 -1 - - 1 1 1 2 2 2 3 3

Normality test (A-D*): P < 0.001

Normal Quantile Plot

Linear price-demand model for CASES_18PK (1 variable, n=52)
3
2
1
0
-1
-2
-3
-3 -2 -1 0 1 2 3
Normality test (A-D*): P < 0.001

End of Output
 Color Font NoHeaders With P-value
Linear price-
R code: Linear.price-demand.model <- lm(CASES_18PK ~ PRICE_18PK, data = Data)
Here the regression procedure has been used to fit a simple regression model for predicting sales of 18
Confidence from price of 18-packs. A customized model name "Linear price-demand model" has been entered in p
95.0% the default "Model 1".
Std. Coeff. The slope coefficient in this model (-93) is the predicted change in number of 18-packs sold per $1 chan
0.000 price. Despite its high value of R-squared (75%), this model is unsatisfactory for a number of reasons:
-0.866 do not have the same variance for large and small predictions (as indicated by the line fit plot and resid
predicted plot), they are not normally distributed (as indicated by the Anderson-Darling statistic and the
histogram and normal quantile plots), and the predicted value of cases-sold is negative for prices great
$19.50 (as is apparent on the line fit plot).

The teaching notes option was selected when running this analysis. Hover the mouse over a cell with a
to see a pop-up note that explains it. (If you don't see red flags, toggle the "Notes" button on the Regre
ribbon if it is running, or toggle the Review/Show-All-Comments button a couple of times on the Excel r

The "Colors" and "Fonts" options have also been turned on for this sheet via the buttons on the Regress
This causes the t-statistics and standardized coefficients to be color-coded for sign and significance and
have bolder fonts for more significant variables. This feature can be toggled on and off while viewing a
sheet.

MASE lag 1
0.401
 Editable 1234 No Comment Notes 6/6/20 2:10 PM + FACDS414 + Beer_sales_
No preceding m
Linear price- Linear price-demand model following model is Log-log model (#vars=1, n=52, AdjRsq=0.886): CASES_18PK.Ln << PRICE_18PK.L
model for predicting sales of 18-packs
nd model" has been entered in place of

ber of 18-packs sold per $1 change in

ctory for a number of reasons: the errors
ted by the line fit plot and residual-vs-
nderson-Darling statistic and the residual
-sold is negative for prices greater than

ver the mouse over a cell with a red flag

the "Notes" button on the RegressIt
a couple of times on the Excel ribbon.)

et via the buttons on the RegressIt ribbon.

ed for sign and significance and also to
gled on and off while viewing a model
M + FACDS414 + Beer_sales_2.xlsx + Data + RegressItPC 2020.03.04
S_18PK.Ln << PRICE_18PK.Ln
6/6/20 2:10 PM + FACDS414 + Beer_sales_2.xlsx + Data + RegressItPC 2020.03.04
Model: Log-log model
Dependent Variable: CASES_18PK.Ln

R-Squared Adj.R-Sqr. Std.Err.Reg. Std.Dep.Var. # Fitted # Missing Critical t

0.888 0.886 0.356 1.054 52 0 2.009

Variable Coefficient Std.Err. t-Statistic P-value Lower95% Upper95% VIF

Constant 23.831 0.946 25.201 0.000 21.932 25.731 0.000
PRICE_18PK.Ln -6.705 0.337 -19.926 0.000 -7.381 -6.029 1.000

PRICE_18PK.LnStdErrMean StdErrFcst Predicted Lower 95% Upper 95%

2.585Log-log model
0.089for CASES_18PK.Ln
0.367 (1 variable,
6.500 n=52) 5.763 7.237
2.681Predicted 0.065
CASES_18PK.Ln0.362
= 23.831 - 6.705*PRICE_18PK.Ln
5.853 5.127 6.580
7.5 2.778 0.050 0.359 5.207 4.485 5.929
7 2.874 0.054 0.360 4.560 3.837 5.284
2.970 0.074 0.363 3.914 3.184 4.644
6.5
6
5.5 Upper 95%
5 Predicted
4.5 Lower 95%
4
3.5
3
2.55 2.6 2.65 2.7 2.75 2.8 2.85 2.9 2.95 3
PRICE_18PK.Ln

Mean Error RMSE MAE Minimum Maximum MAPE A-D* stat

Fitted (n=52) 0.000 0.349 0.281 -0.899 0.595 5.8% 0.34 (P=0.508)

Lag 1 2 3 4
Autocorrelation 0.092 0.087 -0.294 -0.137
StdErrorsFromZero 0.657 0.613 -2.059 -0.952
Durbin-Watson 1.772

Actual and Predicted

1.000-vs- Observation #
Log-log model for CASES_18PK.Ln (1 variable, n=52)
7
6.5
6
5.5
5
4.5
4
3.5
3
0 10 20 30 40 50 60
 Residual -vs- Observation #
Log-log model for CASES_18PK.Ln (1 variable, n=52)
0.8
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
-1
Lag 1 autocorrelation = 0.09

Residual -vs- Predicted

Log-log model for CASES_18PK.Ln (1 variable, n=52)
0.8
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
-1
3.5 4 4.5 5 5.5 6 6.5 7

Histogram of Residuals
Log-log model for CASES_18PK.Ln (1 variable, n=52)
10
8
6
4 Actual
Normal
2
0
00 10 20 30 40 50 60 70 80 90 0 0 90 8 0 7 0 60 5 0 40 3 0 20 1 0 00
.9 .8 .7 .6 .5 .4 .3 .2 .1 .0 .0 .0 .1 .2 .3 .4 .5 .6 .7 .8 .9
-0 -0 -0 -0 -0 -0 -0 -0 -0 -0 0 0 0 0 0 0 0 0 0 0 0

Normality test (A-D*): P > 0.05

 2
0
00 10 20 30 40 50 60 70 80 90 0 0 90 8 0 7 0 60 5 0 40 3 0 20 1 0 00
0 .9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
- - - - - - - - - -

Normality test (A-D*): P > 0.05

Normal Quantile Plot

Log-log model for CASES_18PK.Ln (1 variable, n=52)
3
2
1
0
-1
-2
-3
-3 -2 -1 0 1 2 3
Normality test (A-D*): P > 0.05

End of Output
 Color Font NoHeaders With P-value
The natural log transformation is often used in modeling Log-log model
price-demandrelationships. The
R code: Log-log.model <- lm(CASES_18PK.Ln
price ~ PRICE_18PK.Ln,
can be interpreted data =percent
as the predicted Data)
change in sales per percent change in pr
Confidence demand", and the model assumes it to be the same at all price levels.
95.0%
Here is the result of fitting such a model to sales of 18-packs. The variable transformation
Std. Coeff. transformation to both variables, and a simple regression model was then fitted. (The log
0.000 their names.) The coefficient of log price is -6.7, which means that a 1% decrease in price
-0.942 percentages, the model predicts that this effect will be compounded rather than scaled up

The standard error of the regression in a model fitted to a logged variable is approximatel
errors if it is not too large, say, on the order of 0.10, meaning that the standard deviation o
error of the regression in this model is 0.356, which is too large for that rule to apply very
the percentage errors is in the general vicinity of 36%. (Actually it is 40%.)

The various plots and diagnostic statistics look much better than those of the previous mo
regression. The errors are normally distributed and the variance is roughly the same for la

The forecasts produced by this model can be converted back into real units of cases-sold b
them. In order to be able to do that, the residuals and predictions have been saved back
this model are strictly positive numbers at all price levels, unlike those of the original linea

MASE lag 1
0.319
 Editable 1234 No Comment Notes 6/6/20 2:29 PM + FACDS414 + Beer_sales_
No preceding m
Log-log modelLog-log model following model is Log-log 3-variable model
rice-demand relationships. The slope coefficient in a simple regression of log sales on log (#vars=3, n=52, AdjRsq=0.929): CASES_18PK.Ln << PRICE_12PK.Ln,
n sales per percent change in price, on the margin, which is the so-called "price elasticity of
price levels.

ks. The variable transformation tool in RegressIt was used to apply the natural log
model was then fitted. (The logged variables are identified by the tag ".Ln" at the end of
eans that a 1% decrease in price is predicted to yield a 6.7% increase in sales. For large
mpounded rather than scaled up linearly.

logged variable is approximately the standard deviation of the model's real percentage
ing that the standard deviation of the percentage errors is approximately 10% The standard
large for that rule to apply very accurately, but it indicates that the standard deviation of
tually it is 40%.)

r than those of the previous model in terms of satisfying the assumptions of linear
riance is roughly the same for large and small predictions.

ack into real units of cases-sold by "unlogging" them, i.e., applying the EXP function to
edictions have been saved back to the data sheet at run time. The unlogged forecasts from
unlike those of the original linear model.
M + FACDS414 + Beer_sales_6.xlsx + Data + RegressItPC 2020.03.04
18PK.Ln << PRICE_12PK.Ln, PRICE_18PK.Ln, PRICE_30PK.Ln
6/6/20 2:29 PM + FACDS414 + Beer_sales_6.xlsx + Data + RegressItPC 2020.03.04
Model: Log-log 3-variable model
Dependent Variable: CASES_18PK.Ln

R-Squared Adj.R-Sqr. Std.Err.Reg. Std.Dep.Var. # Fitted # Missing Critical t

0.933 0.929 0.281 1.054 52 0 2.011

Variable Coefficient Std.Err. t-Statistic P-value Lower95% Upper95% VIF

Constant 10.267 3.024 3.395 0.001 4.187 16.347 0.000
PRICE_12PK.Ln 2.016 0.360 5.608 0.000 1.293 2.739 1.196
PRICE_18PK.Ln -6.331 0.280 -22.647 0.000 -6.893 -5.769 1.111
PRICE_30PK.Ln 2.470 0.781 3.161 0.003 0.899 4.041 1.270

Mean Error RMSE MAE Minimum Maximum MAPE A-D* stat

Fitted (n=52) 0.000 0.270 0.204 -0.742 0.681 4.2% 0.66 (P=0.087)

Lag 1 2 3 4
Autocorrelation 0.180 0.278 0.079 0.095
StdErrorsFromZero 1.286 1.967 0.552 0.657
Durbin-Watson 1.596

Actual and Predicted -vs- Observation #

1.000
Log-log 3-variable model for CASES_18PK.Ln (3 variables, n=52)
7
6.5
6
5.5
5
4.5
4
3.5
3
0 10 20 30 40 50 60

Residual -vs- Observation #

Log-log 3-variable model for CASES_18PK.Ln (3 variables, n=52)
0.8
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
-1
Lag 1 autocorrelation = 0.18
-0.4
-0.6
-0.8
-1
Lag 1 autocorrelation = 0.18

Residual -vs- Predicted

Log-log 3-variable model for CASES_18PK.Ln (3 variables, n=52)
0.8
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
-1
3.5 4 4.5 5 5.5 6 6.5 7

Histogram of Residuals
Log-log 3-variable model for CASES_18PK.Ln (3 variables, n=52)
10
8
6
4 Actual
Normal
2
0
50 75 00 25 50 75 00 25 50 75 0 0 75 5 0 2 5 00 7 5 50 2 5 00 7 5 50
0 .7 0.6 0.6 0.5 0.4 0.3 0.3 0.2 0.1 0.0 0.0 0.0 0.1 0.2 0.3 0.3 0.4 0.5 0.6 0.6 0.7
- - - - - - - - - -

Normality test (A-D*): P > 0.05

Normal Quantile Plot

Log-log 3-variable model for CASES_18PK.Ln (3 variables, n=52)
3
2
1
0
-1
-2
-3
-3 -2 -1 0 1 2 3
Normality test (A-D*): P > 0.05
 0
-1
-2
-3
-3 -2 -1 0 1 2 3
Normality test (A-D*): P > 0.05

End of Output
 Color Font NoHeaders With P-value
Log-log 3-var
R code: Log-log.3-variable.model <- lm(CASES_18PK.Ln ~ PRICE_12PK.Ln + PRICE_18PK.Ln + PRICE_30PK.Ln, data = Data)
Confidence
The addition of the logged prices of other two carton sizes yields a significant
95.0% reduction in the standard error of the regression compared to the previous model
(from 0.356 down to 0.281), as well as a significant further reduction in the error
Std. Coeff.
measures in real terms. The coefficients of the logged prices of the other two size
0.000
cartons are "cross-price elasticities." However, a slight time trend now appears in
0.229
the errors.
-0.890
Notice that the cells for t-stats and standardized coefficients have blue fill color when
0.133 they are positive, as they are here for the two new variables.

MASE lag 1
0.232
 Editable 1234 No Comment Notes 6/6/20 2:46 PM + FACDS414 + Beer_sales_
Log-log 3-var Log-log 3-var Log-log 3-variable model following model is Log-log 4-variable model (#vars=4, n=52, AdjRsq=0.947): CASES_18PK.Ln << PRICE
Ln + PRICE_30PK.Ln, data = Data)
ields a significant
o the previous model
eduction in the error
of the other two size
rend now appears in

have blue fill color when

M + FACDS414 + Beer_sales_11.xlsx + Data + RegressItPC 2020.03.04
): CASES_18PK.Ln << PRICE_12PK.Ln, PRICE_18PK.Ln, PRICE_30PK.Ln, Week
6/6/20 2:46 PM + FACDS414 + Beer_sales_11.xlsx + Data + RegressItPC 2020.03.04
Model: Log-log 4-variable model
Dependent Variable: CASES_18PK.Ln

R-Squared Adj.R-Sqr. Std.Err.Reg. Std.Dep.Var. # Fitted # Missing Critical t

0.951 0.947 0.244 1.054 52 0 2.012

Variable Coefficient Std.Err. t-Statistic P-value Lower95% Upper95% VIF

Constant 8.275 2.671 3.098 0.003 2.901 13.648 0.000
PRICE_12PK.Ln 2.524 0.336 7.508 0.000 1.847 3.200 1.386
PRICE_18PK.Ln -5.896 0.265 -22.240 0.000 -6.430 -5.363 1.325
PRICE_30PK.Ln 2.089 0.685 3.049 0.004 0.711 3.467 1.294
Week 0.011 0.002738 4.079 0.000 0.005660 0.017 1.480

Mean Error RMSE MAE Minimum Maximum MAPE A-D* stat

Fitted (n=52) 0.000 0.232 0.179 -0.642 0.568 3.7% 0.25 (P=0.738)

Lag 1 2 3 4
Autocorrelation 0.078 0.103 0.033 -0.017
StdErrorsFromZero 0.557 0.726 0.233 -0.121
Durbin-Watson 1.827

Actual and Predicted -vs- Observation #

1.000
Log-log 4-variable model for CASES_18PK.Ln (4 variables, n=52)
7.5
7
6.5
6
5.5
5
4.5
4
3.5
3
0 10 20 30 40 50 60

Residual -vs- Observation #

Log-log 4-variable model for CASES_18PK.Ln (4 variables, n=52)
0.8
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
Lag 1 autocorrelation = 0.08
-0.2
-0.4
-0.6
-0.8
Lag 1 autocorrelation = 0.08

Residual -vs- Predicted

Log-log 4-variable model for CASES_18PK.Ln (4 variables, n=52)
0.8
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
3.5 4 4.5 5 5.5 6 6.5 7 7.5

Histogram of Residuals
Log-log 4-variable model for CASES_18PK.Ln (4 variables, n=52)
12
10
8
6
Actual
4
Normal
2
0
50 85 20 55 90 25 60 95 30 65 0 0 65 3 0 9 5 60 2 5 90 5 5 20 8 5 50
0 .6 0.5 0.5 0.4 0.3 0.3 0.2 0.1 0.1 0.0 0.0 0.0 0.1 0.1 0.2 0.3 0.3 0.4 0.5 0.5 0.6
- - - - - - - - - -

Normality test (A-D*): P > 0.05

Normal Quantile Plot

Log-log 4-variable model for CASES_18PK.Ln (4 variables, n=52)
3
2
1
0
-1
-2
-3
-4
-3 -2 -1 0 1 2 3
Normality test (A-D*): P > 0.05
 0
-1
-2
-3
-4
-3 -2 -1 0 1 2 3
Normality test (A-D*): P > 0.05

End of Output
 Color Font NoHeaders With P-value
Log-log 4-var
R code: Log-log.4-variable.model <- lm(CASES_18PK.Ln ~ PRICE_12PK.Ln + PRICE_18PK.Ln + PRICE_30PK.Ln + Week, data =
Confidence
95.0%
Addition of the time index (Week) as a fourth
Std. Coeff. variable eliminates the time trend in the errors
0.000 and yields a slight further reduction in the
0.286 standard error of the regression, from 0.281
-0.829 down to 0.244.
0.112
0.161

MASE lag 1
0.204
 Editable 1234 No Comment Notes 6/6/20 2:49 PM + FACDS414 + Beer_sales_
Log-log 4-var No following model in this sequence.
Ln + PRICE_30PK.Ln + Week, data = Data)
M + FACDS414 + Beer_sales_13.xlsx + Data + RegressItPC 2020.03.04
6/6/20 2:49 PM + FACDS414 + Beer_sales_13.xlsx + Data + RegressItPC 2020.03.04
 Summary of Regression Model Results
ce-demand model (#vars=1, n=52, AdjRsq=0.746): CASES_18PK << PRICE_18PK
Linear Model For CASES_18PK Linear price-demand model
Run Time 6/6/20 2:10 PM
# Fitted 52
Mean 256.673
Standard Deviation 258.825
Number Of Variables 1
Standard Error of Regression 130.529
R-squared 0.751
Adjusted R-squared 0.746
Mean Absolute Error 82.495
Mean Absolute Percentage Error 42.8%
Maximum VIF
Normality Test ***
Mean Absolute Scaled Error 0.401 (lag 1)
Residual Autocorrelation -0.22 (lag 1)

Coefficients: Linear price-demand model

Constant 1,812 (0.000)
PRICE_18PK -93.007 (0.000)

log model (#vars=1, n=52, AdjRsq=0.886): CASES_18PK.Ln << PRICE_18PK.Ln RICE_18PK.Ln, PRICE_30PK.Ln 8PK.Ln, PRICE_30PK.Ln, Week
Linear Model For CASES_18PK.Ln Log-log model Log-log 3-variable model Log-log 4-variable model
Run Time 6/6/20 2:29 PM 6/6/20 2:46 PM 6/6/20 2:49 PM
# Fitted 52 52 52
Mean 5.014 5.014 5.014
Standard Deviation 1.054 1.054 1.054
Number Of Variables 1 3 4
Standard Error of Regression 0.356 0.281 0.244
R-squared 0.888 0.933 0.951
Adjusted R-squared 0.886 0.929 0.947
Mean Absolute Error 0.281 0.204 0.179
Mean Absolute Percentage Error 5.8% 4.2% 3.7%
Maximum VIF 1.270 1.480
Normality Test _ _ _
Mean Absolute Scaled Error 0.319 (lag 1) 0.232 (lag 1) 0.204 (lag 1)
Residual Autocorrelation 0.09 (lag 1) 0.18 (lag 1) 0.08 (lag 1)

Coefficients: Log-log model Log-log 3-variable model Log-log 4-variable model

Constant 23.831 (0.000) 10.267 (0.001) 8.275 (0.003)
PRICE_12PK.Ln 2.016 (0.000) 2.524 (0.000)
PRICE_18PK.Ln -6.705 (0.000) -6.331 (0.000) -5.896 (0.000)
PRICE_30PK.Ln 2.470 (0.003) 2.089 (0.004)
Week 0.011 (0.000)
 Color Font

The model summary worksheet provides an audit trail of all

regression models fitted in the same workbook, showing the run
time, variable selections, coefficient estimates, and error stats for
each one.

The results for models that used the same dependent variable are
arranged side-by-side.

More audit trail information about the model appears in a cell

comment, as on the model sheet.

The analysis history list (launched from the History button on the
ribbon) provides a more compact view of the current contents of
the workbook, as in the screen shot below.
Comment
Week Date PRICE 12PKPRICE 18PKPRICE 30PKCASES 12PCASES 18PCASES 30PK
1 1/1/2003 19.98 14.10 15.19 223.5 439.0 55.00
2 1/8/2003 19.98 18.65 15.19 215.0 98.0 66.75 Here's the original data prior to
creation of transformed (logged
3 1/15/2003 19.98 18.65 13.87 227.5 70.0 242.00 variables.
4 1/22/2003 19.98 18.65 12.83 244.5 52.0 488.50
5 1/29/2003 19.98 18.65 13.16 313.5 64.0 308.75
6 2/5/2003 19.98 18.65 15.19 279.0 72.0 111.75
7 2/12/2003 19.98 18.65 13.92 238.0 47.0 252.50
8 2/19/2003 20.10 18.73 14.42 315.5 85.0 221.25
9 2/26/2003 20.12 18.75 13.83 217.0 59.0 245.25
10 3/5/2003 20.13 18.75 14.50 209.5 63.0 148.50
11 3/12/2003 20.14 18.75 13.87 227.0 57.0 229.75
12 3/19/2003 20.12 18.75 13.64 216.5 54.0 312.00
13 3/26/2003 20.12 13.87 14.31 169.0 404.0 96.75
14 4/2/2003 20.13 14.27 13.85 178.0 380.0 123.25
15 4/9/2003 20.14 18.76 14.20 301.5 65.0 200.50
16 4/16/2003 20.14 18.77 13.64 266.5 40.0 359.75
17 4/23/2003 20.13 13.87 14.33 182.5 456.0 113.50
18 4/30/2003 20.13 14.14 13.14 159.0 176.0 136.50
19 5/7/2003 20.13 18.76 13.81 285.5 61.0 225.50
20 5/14/2003 20.13 18.72 15.19 360.0 91.0 122.25
21 5/21/2003 20.13 18.76 13.13 263.0 59.0 443.75
22 5/28/2003 19.18 18.76 13.63 443.5 83.0 322.75
23 6/4/2003 14.78 18.74 15.19 1101.5 41.0 53.00
24 6/11/2003 16.04 18.75 13.89 814.0 47.0 140.75
25 6/18/2003 20.12 18.75 14.28 365.0 84.0 210.75
26 6/25/2003 19.75 18.75 15.19 510.0 85.0 110.50
27 7/2/2003 19.65 18.75 13.12 580.5 116.0 568.25
28 7/9/2003 19.69 13.79 13.78 251.0 544.0 115.50
29 7/16/2003 20.12 13.49 15.19 237.0 890.0 58.75
30 7/23/2003 20.12 14.89 15.19 302.5 371.0 77.25
31 7/30/2003 20.13 13.94 15.19 229.5 557.0 66.25
32 8/6/2003 20.14 13.67 15.19 188.5 775.0 50.00
33 8/13/2003 15.14 14.43 15.19 795.5 236.0 46.50
34 8/20/2003 14.33 18.75 15.19 1556.5 43.0 65.75
35 8/27/2003 16.24 18.22 13.14 807.5 63.0 252.75
36 9/3/2003 19.93 14.06 13.45 243.0 469.0 179.00
37 9/10/2003 21.06 14.43 13.00 201.5 335.0 226.25
38 9/17/2003 21.19 19.48 13.60 294.0 75.0 288.50
39 9/24/2003 21.23 15.15 14.46 220.5 461.0 114.25
40 10/1/2003 20.12 13.79 14.94 255.5 817.0 70.00
41 10/8/2003 14.73 14.31 15.19 920.5 200.0 47.75
42 10/15/2003 14.57 19.50 15.19 730.0 32.0 98.75
43 10/22/2003 15.94 13.85 15.19 262.5 460.0 77.00
44 10/29/2003 20.70 14.23 13.43 209.5 751.0 160.50
45 11/5/2003 19.57 19.31 14.37 283.0 70.0 143.50
46 11/12/2003 19.60 19.29 15.19 262.5 80.0 133.00
47 11/19/2003 19.94 13.76 15.19 310.0 523.0 68.75
48 11/26/2003 21.28 13.45 15.19 278.5 741.0 81.75
49 12/3/2003 14.56 15.13 15.19 741.5 130.0 56.25
50 12/10/2003 14.39 19.43 15.19 1316.0 69.0 68.75
51 12/17/2003 16.81 13.26 15.19 449.0 493.0 49.25
52 12/24/2003 19.86 13.92 15.19 505.0 814.0 76.50
e's the original data prior to the
tion of transformed (logged)
ables.