library(arules)Loading required package: Matrix
Attaching package: 'arules'The following objects are masked from 'package:base':
abbreviate, writeAssignment Version 3
Number 1
retail_transactions <- read.transactions("retail_transactions_3.csv", sep= ",")Warning in scan(text = l, what = "character", sep = sep, quote = quote, : EOF
within quoted string
Warning in scan(text = l, what = "character", sep = sep, quote = quote, : EOF
within quoted string
Warning in scan(text = l, what = "character", sep = sep, quote = quote, : EOF
within quoted string
Warning in scan(text = l, what = "character", sep = sep, quote = quote, : EOF
within quoted string
Warning in scan(text = l, what = "character", sep = sep, quote = quote, : EOF
within quoted string
Warning in scan(text = l, what = "character", sep = sep, quote = quote, : EOF
within quoted string
Warning in scan(text = l, what = "character", sep = sep, quote = quote, : EOF
within quoted string
Warning in scan(text = l, what = "character", sep = sep, quote = quote, : EOF
within quoted stringWarning in asMethod(object): removing duplicated items in transactionsNumber 2
transactions as itemMatrix in sparse format with
10000 rows (elements/itemsets/transactions) and
5479 columns (items) and a density of 0.002744552
most frequent items:
WHITE HANGING HEART T-LIGHT HOLDER REGENCY CAKESTAND 3 TIER
822 776
JUMBO BAG RED RETROSPOT PARTY BUNTING
663 561
ASSORTED COLOUR BIRD ORNAMENT (Other)
544 147008
element (itemset/transaction) length distribution:
sizes
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
1658 707 498 413 365 341 316 310 309 290 261 227 227 242 260 227
17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32
204 199 233 189 184 149 138 125 104 111 112 98 113 99 78 68
33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48
55 62 65 44 48 41 52 41 44 26 45 27 27 35 30 24
49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64
26 23 22 21 16 19 21 13 11 16 13 14 15 10 13 13
65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80
3 13 13 7 6 9 9 6 6 4 3 8 5 6 3 5
81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96
3 4 5 8 3 5 8 2 4 4 1 3 2 3 1 2
97 98 100 101 102 103 104 105 107 108 109 110 111 112 113 119
5 1 2 2 2 1 2 1 1 3 1 2 1 1 2 1
120 121 122 123 125 127 134 142 146 147 150 154 157 171 193 204
1 1 1 1 1 1 1 2 1 1 1 2 1 2 1 1
235 249
1 1
Min. 1st Qu. Median Mean 3rd Qu. Max.
1.00 3.00 10.00 15.04 21.00 249.00
includes extended item information - examples:
labels
1 1 HANGER
2 10 COLOUR SPACEBOY PEN
3 12 COLOURED PARTY BALLOONS2(a)
There was 10,000 transactions within the dataset
2(b)
There was 5,479 items available for purchase
2(c)
Number of rows: 10,000
Number of columns: 5,479
Density: 0.002744552
The sparse matrix contains a total of 54,790,000 cells (10,000 rows * 5,479 columns).
150,553 cells contain non-zero values.
2(d)
The largest number of items purchased in a single transaction is 249
2(e)
The mean number of items purchased in a single transaction is 15.04
Number 3
Number 4
Apriori
Parameter specification:
confidence minval smax arem aval originalSupport maxtime support minlen
0.5 0.1 1 none FALSE TRUE 5 0.01 2
maxlen target ext
10 rules TRUE
Algorithmic control:
filter tree heap memopt load sort verbose
0.1 TRUE TRUE FALSE TRUE 2 TRUE
Absolute minimum support count: 100
set item appearances ...[0 item(s)] done [0.00s].
set transactions ...[5479 item(s), 10000 transaction(s)] done [0.04s].
sorting and recoding items ... [384 item(s)] done [0.00s].
creating transaction tree ... done [0.00s].
checking subsets of size 1 2 3 4 done [0.00s].
writing ... [86 rule(s)] done [0.00s].
creating S4 object ... done [0.00s].retail_rulesset of 86 rules 4(a)
This discovered 86 rules using a support threshold of 0.01, a confidence threshold of 0.5 and a minimum length of 2.
4(b)
A support threshold of 0.01 means the algorithm will only consider item-sets that appear together at least 1% of the time. Aiming to focus on item-sets that occur more frequently together.
4(c)
A confidence threshold of 0.5 tells us that the algorithm will only accept rules where there’s a good chance that if someone buys item A, they’ll also purchase item B.
Number 5
set of 86 rules
rule length distribution (lhs + rhs):sizes
2 3
47 39
Min. 1st Qu. Median Mean 3rd Qu. Max.
2.000 2.000 2.000 2.453 3.000 3.000
summary of quality measures:
support confidence coverage lift
Min. :0.01000 Min. :0.5021 Min. :0.01200 Min. : 6.585
1st Qu.:0.01063 1st Qu.:0.5510 1st Qu.:0.01733 1st Qu.:12.347
Median :0.01160 Median :0.6022 Median :0.02000 Median :15.266
Mean :0.01315 Mean :0.6198 Mean :0.02152 Mean :17.144
3rd Qu.:0.01515 3rd Qu.:0.6652 3rd Qu.:0.02428 3rd Qu.:21.864
Max. :0.02270 Max. :0.8814 Max. :0.03770 Max. :44.055
count
Min. :100.0
1st Qu.:106.2
Median :116.0
Mean :131.5
3rd Qu.:151.5
Max. :227.0
mining info:
data ntransactions support confidence
retail_transactions 10000 0.01 0.5
call
apriori(data = retail_transactions, parameter = list(support = 0.01, confidence = 0.5, minlen = 2))5(a)
There are 47 rules with two items
And 39 rules with three items
5(b)
Minimum lift: 6.585
Maximum lift: 44.055
Number 6
6(a)
lhs rhs support confidence coverage lift count
[1] {WOODEN STAR CHRISTMAS SCANDINAVIAN} => {WOODEN HEART CHRISTMAS SCANDINAVIAN} 0.0113 0.7533333 0.0150 44.054581 113
[2] {WOODEN HEART CHRISTMAS SCANDINAVIAN} => {WOODEN STAR CHRISTMAS SCANDINAVIAN} 0.0113 0.6608187 0.0171 44.054581 113
[3] {PINK REGENCY TEACUP AND SAUCER,
ROSES REGENCY TEACUP AND SAUCER} => {GREEN REGENCY TEACUP AND SAUCER} 0.0156 0.8813559 0.0177 28.708662 156
[4] {GREEN REGENCY TEACUP AND SAUCER,
REGENCY CAKESTAND 3 TIER} => {PINK REGENCY TEACUP AND SAUCER} 0.0102 0.6938776 0.0147 28.672626 102
[5] {GREEN REGENCY TEACUP AND SAUCER,
ROSES REGENCY TEACUP AND SAUCER} => {PINK REGENCY TEACUP AND SAUCER} 0.0156 0.6872247 0.0227 28.397714 156
[6] {PINK REGENCY TEACUP AND SAUCER,
REGENCY CAKESTAND 3 TIER} => {GREEN REGENCY TEACUP AND SAUCER} 0.0102 0.8500000 0.0120 27.687296 102
[7] {GARDENERS KNEELING PAD KEEP CALM} => {GARDENERS KNEELING PAD CUP OF TEA} 0.0163 0.5821429 0.0280 26.105061 163
[8] {GARDENERS KNEELING PAD CUP OF TEA} => {GARDENERS KNEELING PAD KEEP CALM} 0.0163 0.7309417 0.0223 26.105061 163
[9] {PINK REGENCY TEACUP AND SAUCER} => {GREEN REGENCY TEACUP AND SAUCER} 0.0192 0.7933884 0.0242 25.843271 192
[10] {GREEN REGENCY TEACUP AND SAUCER} => {PINK REGENCY TEACUP AND SAUCER} 0.0192 0.6254072 0.0307 25.843271 192
[11] {DOLLY GIRL LUNCH BOX} => {SPACEBOY LUNCH BOX} 0.0138 0.6359447 0.0217 25.642931 138
[12] {SPACEBOY LUNCH BOX} => {DOLLY GIRL LUNCH BOX} 0.0138 0.5564516 0.0248 25.642931 138
[13] {GREEN REGENCY TEACUP AND SAUCER,
PINK REGENCY TEACUP AND SAUCER} => {ROSES REGENCY TEACUP AND SAUCER} 0.0156 0.8125000 0.0192 24.399399 156
[14] {REGENCY CAKESTAND 3 TIER,
ROSES REGENCY TEACUP AND SAUCER} => {GREEN REGENCY TEACUP AND SAUCER} 0.0116 0.7483871 0.0155 24.377430 116
[15] {JUMBO BAG PEARS} => {JUMBO BAG APPLES} 0.0111 0.6529412 0.0170 24.272906 111
[16] {GREEN REGENCY TEACUP AND SAUCER,
REGENCY CAKESTAND 3 TIER} => {ROSES REGENCY TEACUP AND SAUCER} 0.0116 0.7891156 0.0147 23.697167 116
[17] {GREEN REGENCY TEACUP AND SAUCER} => {ROSES REGENCY TEACUP AND SAUCER} 0.0227 0.7394137 0.0307 22.204615 227
[18] {ROSES REGENCY TEACUP AND SAUCER} => {GREEN REGENCY TEACUP AND SAUCER} 0.0227 0.6816817 0.0333 22.204615 227
[19] {ROUND SNACK BOXES SET OF 4 FRUITS} => {ROUND SNACK BOXES SET OF4 WOODLAND} 0.0100 0.5524862 0.0181 22.188200 100
[20] {PINK REGENCY TEACUP AND SAUCER} => {ROSES REGENCY TEACUP AND SAUCER} 0.0177 0.7314050 0.0242 21.964113 177
[21] {ROSES REGENCY TEACUP AND SAUCER} => {PINK REGENCY TEACUP AND SAUCER} 0.0177 0.5315315 0.0333 21.964113 177
[22] {ALARM CLOCK BAKELIKE GREEN,
ALARM CLOCK BAKELIKE PINK} => {ALARM CLOCK BAKELIKE RED} 0.0100 0.7575758 0.0132 21.895253 100
[23] {LARGE WHITE HEART OF WICKER} => {SMALL WHITE HEART OF WICKER} 0.0113 0.5159817 0.0219 21.771381 113
[24] {ALARM CLOCK BAKELIKE PINK,
ALARM CLOCK BAKELIKE RED} => {ALARM CLOCK BAKELIKE GREEN} 0.0100 0.6666667 0.0150 21.299255 100
[25] {CHARLOTTE BAG PINK POLKADOT} => {RED RETROSPOT CHARLOTTE BAG} 0.0127 0.6256158 0.0203 20.923604 127
[26] {HOT WATER BOTTLE I AM SO POORLY} => {CHOCOLATE HOT WATER BOTTLE} 0.0102 0.5454545 0.0187 20.661157 102
[27] {STRAWBERRY CHARLOTTE BAG} => {RED RETROSPOT CHARLOTTE BAG} 0.0117 0.6157895 0.0190 20.594966 117
[28] {ALARM CLOCK BAKELIKE GREEN,
ALARM CLOCK BAKELIKE RED} => {ALARM CLOCK BAKELIKE PINK} 0.0100 0.5025126 0.0199 20.344638 100
[29] {ALARM CLOCK BAKELIKE IVORY} => {ALARM CLOCK BAKELIKE RED} 0.0120 0.6896552 0.0174 19.932230 120
[30] {BAKING SET SPACEBOY DESIGN} => {BAKING SET 9 PIECE RETROSPOT} 0.0115 0.6388889 0.0180 19.597819 115
[31] {ALARM CLOCK BAKELIKE GREEN} => {ALARM CLOCK BAKELIKE RED} 0.0199 0.6357827 0.0313 18.375224 199
[32] {ALARM CLOCK BAKELIKE RED} => {ALARM CLOCK BAKELIKE GREEN} 0.0199 0.5751445 0.0346 18.375224 199
[33] {LUNCH BAG CARS BLUE,
LUNCH BAG RED RETROSPOT} => {LUNCH BAG WOODLAND} 0.0101 0.5343915 0.0189 17.578669 101
[34] {ALARM CLOCK BAKELIKE PINK} => {ALARM CLOCK BAKELIKE RED} 0.0150 0.6072874 0.0247 17.551660 150
[35] {CHARLOTTE BAG SUKI DESIGN} => {RED RETROSPOT CHARLOTTE BAG} 0.0114 0.5181818 0.0220 17.330496 114
[36] {WOODLAND CHARLOTTE BAG} => {RED RETROSPOT CHARLOTTE BAG} 0.0111 0.5115207 0.0217 17.107717 111
[37] {ALARM CLOCK BAKELIKE PINK} => {ALARM CLOCK BAKELIKE GREEN} 0.0132 0.5344130 0.0247 17.073896 132
[38] {WOODEN PICTURE FRAME WHITE FINISH} => {WOODEN FRAME ANTIQUE WHITE} 0.0197 0.5487465 0.0359 16.283279 197
[39] {WOODEN FRAME ANTIQUE WHITE} => {WOODEN PICTURE FRAME WHITE FINISH} 0.0197 0.5845697 0.0337 16.283279 197
[40] {LUNCH BAG CARS BLUE,
LUNCH BAG RED RETROSPOT} => {LUNCH BAG PINK POLKADOT} 0.0116 0.6137566 0.0189 16.280016 116
[41] {LUNCH BAG BLACK SKULL,
LUNCH BAG CARS BLUE} => {LUNCH BAG PINK POLKADOT} 0.0108 0.6000000 0.0180 15.915119 108
[42] {LUNCH BAG RED RETROSPOT,
LUNCH BAG SUKI DESIGN} => {LUNCH BAG PINK POLKADOT} 0.0103 0.5953757 0.0173 15.792459 103
[43] {LUNCH BAG RED RETROSPOT,
LUNCH BAG WOODLAND} => {LUNCH BAG CARS BLUE} 0.0101 0.5906433 0.0171 15.301639 101
[44] {LUNCH BAG BLACK SKULL,
LUNCH BAG RED RETROSPOT} => {LUNCH BAG PINK POLKADOT} 0.0120 0.5741627 0.0209 15.229779 120
[45] {LUNCH BAG DOLLY GIRL DESIGN} => {LUNCH BAG SPACEBOY DESIGN} 0.0131 0.5796460 0.0226 15.213806 131
[46] {LUNCH BAG BLACK SKULL,
LUNCH BAG PINK POLKADOT} => {LUNCH BAG CARS BLUE} 0.0108 0.5775401 0.0187 14.962179 108
[47] {LUNCH BAG CARS BLUE,
LUNCH BAG RED RETROSPOT} => {LUNCH BAG BLACK SKULL} 0.0118 0.6243386 0.0189 14.829896 118
[48] {LUNCH BAG RED RETROSPOT,
LUNCH BAG SPACEBOY DESIGN} => {LUNCH BAG PINK POLKADOT} 0.0101 0.5580110 0.0181 14.801354 101
[49] {LUNCH BAG PINK POLKADOT,
LUNCH BAG RED RETROSPOT} => {LUNCH BAG CARS BLUE} 0.0116 0.5686275 0.0204 14.731281 116
[50] {LUNCH BAG BLACK SKULL,
LUNCH BAG RED RETROSPOT} => {LUNCH BAG CARS BLUE} 0.0118 0.5645933 0.0209 14.626769 118
[51] {LUNCH BAG CARS BLUE,
LUNCH BAG PINK POLKADOT} => {LUNCH BAG BLACK SKULL} 0.0108 0.6101695 0.0177 14.493337 108
[52] {LUNCH BAG VINTAGE LEAF DESIGN} => {LUNCH BAG APPLE DESIGN} 0.0122 0.5020576 0.0243 14.468519 122
[53] {LUNCH BAG RED RETROSPOT,
LUNCH BAG WOODLAND} => {LUNCH BAG BLACK SKULL} 0.0102 0.5964912 0.0171 14.168438 102
[54] {LUNCH BAG PINK POLKADOT,
LUNCH BAG RED RETROSPOT} => {LUNCH BAG SUKI DESIGN} 0.0103 0.5049020 0.0204 14.064121 103
[55] {LUNCH BAG RED RETROSPOT,
LUNCH BAG SPACEBOY DESIGN} => {LUNCH BAG BLACK SKULL} 0.0107 0.5911602 0.0181 14.041810 107
[56] {LUNCH BAG RED RETROSPOT,
LUNCH BAG SUKI DESIGN} => {LUNCH BAG BLACK SKULL} 0.0102 0.5895954 0.0173 14.004641 102
[57] {LUNCH BAG PINK POLKADOT,
LUNCH BAG RED RETROSPOT} => {LUNCH BAG BLACK SKULL} 0.0120 0.5882353 0.0204 13.972335 120
[58] {LUNCH BAG CARS BLUE,
LUNCH BAG WOODLAND} => {LUNCH BAG RED RETROSPOT} 0.0101 0.7266187 0.0139 13.735703 101
[59] {LUNCH BAG BLACK SKULL,
LUNCH BAG WOODLAND} => {LUNCH BAG RED RETROSPOT} 0.0102 0.7183099 0.0142 13.578636 102
[60] {LUNCH BAG BLACK SKULL,
LUNCH BAG RED RETROSPOT} => {LUNCH BAG SPACEBOY DESIGN} 0.0107 0.5119617 0.0209 13.437316 107
[61] {LUNCH BAG PINK POLKADOT,
LUNCH BAG SPACEBOY DESIGN} => {LUNCH BAG RED RETROSPOT} 0.0101 0.7013889 0.0144 13.258769 101
[62] {PAINTED METAL PEARS ASSORTED} => {ASSORTED COLOUR BIRD ORNAMENT} 0.0106 0.7210884 0.0147 13.255302 106
[63] {LUNCH BAG BLACK SKULL,
LUNCH BAG CARS BLUE} => {LUNCH BAG RED RETROSPOT} 0.0118 0.6555556 0.0180 12.392355 118
[64] {LUNCH BAG CARS BLUE,
LUNCH BAG PINK POLKADOT} => {LUNCH BAG RED RETROSPOT} 0.0116 0.6553672 0.0177 12.388795 116
[65] {LUNCH BAG BLACK SKULL,
LUNCH BAG SPACEBOY DESIGN} => {LUNCH BAG RED RETROSPOT} 0.0107 0.6524390 0.0164 12.333441 107
[66] {LUNCH BAG PINK POLKADOT,
LUNCH BAG SUKI DESIGN} => {LUNCH BAG RED RETROSPOT} 0.0103 0.6518987 0.0158 12.323227 103
[67] {60 TEATIME FAIRY CAKE CASES} => {PACK OF 72 RETROSPOT CAKE CASES} 0.0134 0.5056604 0.0265 12.184587 134
[68] {LUNCH BAG BLACK SKULL,
LUNCH BAG PINK POLKADOT} => {LUNCH BAG RED RETROSPOT} 0.0120 0.6417112 0.0187 12.130647 120
[69] {LUNCH BAG BLACK SKULL,
LUNCH BAG SUKI DESIGN} => {LUNCH BAG RED RETROSPOT} 0.0102 0.6144578 0.0166 11.615460 102
[70] {LUNCH BAG WOODLAND} => {LUNCH BAG RED RETROSPOT} 0.0171 0.5625000 0.0304 10.633270 171
[71] {LUNCH BAG PINK POLKADOT} => {LUNCH BAG RED RETROSPOT} 0.0204 0.5411141 0.0377 10.228999 204
[72] {LUNCH BAG DOLLY GIRL DESIGN} => {LUNCH BAG RED RETROSPOT} 0.0121 0.5353982 0.0226 10.120950 121
[73] {JUMBO BAG STRAWBERRY} => {JUMBO BAG RED RETROSPOT} 0.0172 0.6515152 0.0264 9.826775 172
[74] {JUMBO BAG SCANDINAVIAN BLUE PAISLEY} => {JUMBO BAG RED RETROSPOT} 0.0108 0.6352941 0.0170 9.582113 108
[75] {JUMBO BAG PINK POLKADOT} => {JUMBO BAG RED RETROSPOT} 0.0223 0.6043360 0.0369 9.115174 223
[76] {CANDLEHOLDER PINK HANGING HEART} => {WHITE HANGING HEART T-LIGHT HOLDER} 0.0111 0.7449664 0.0149 9.062852 111
[77] {JUMBO BAG SPACEBOY DESIGN} => {JUMBO BAG RED RETROSPOT} 0.0113 0.5978836 0.0189 9.017852 113
[78] {JUMBO STORAGE BAG SUKI} => {JUMBO BAG RED RETROSPOT} 0.0186 0.5904762 0.0315 8.906127 186
[79] {RED HANGING HEART T-LIGHT HOLDER} => {WHITE HANGING HEART T-LIGHT HOLDER} 0.0186 0.6888889 0.0270 8.380643 186
[80] {JUMBO BAG BAROQUE BLACK WHITE} => {JUMBO BAG RED RETROSPOT} 0.0147 0.5505618 0.0267 8.304100 147
[81] {JUMBO BAG PINK VINTAGE PAISLEY} => {JUMBO BAG RED RETROSPOT} 0.0128 0.5493562 0.0233 8.285916 128
[82] {JUMBO BAG VINTAGE DOILY} => {JUMBO BAG RED RETROSPOT} 0.0116 0.5155556 0.0225 7.776102 116
[83] {JUMBO SHOPPER VINTAGE RED PAISLEY} => {JUMBO BAG RED RETROSPOT} 0.0152 0.5033113 0.0302 7.591422 152
[84] {JUMBO BAG WOODLAND ANIMALS} => {JUMBO BAG RED RETROSPOT} 0.0101 0.5024876 0.0201 7.578998 101
[85] {GREEN REGENCY TEACUP AND SAUCER,
PINK REGENCY TEACUP AND SAUCER} => {REGENCY CAKESTAND 3 TIER} 0.0102 0.5312500 0.0192 6.846005 102
[86] {GREEN REGENCY TEACUP AND SAUCER,
ROSES REGENCY TEACUP AND SAUCER} => {REGENCY CAKESTAND 3 TIER} 0.0116 0.5110132 0.0227 6.585222 1166(i)
The rule “{WOODEN STAR CHRISTMAS SCANDINAVIAN} => {WOODEN HEART CHRISTMAS SCANDINAVIAN}” tells us that customers who purchase “Wooden Star Christmas Scandinavian” are also very likely to purchase “Wooden Heart Christmas Scandinavian”.
6(ii)
With a support of 0.0113, it tells us that 1.13% of all transactions include both “Wooden Star Christmas Scandinavian” and “Wooden Heart Christmas Scandinavian”.
While with a confidence value of 0.75, we know that 75% of all transactions that contain specifically “Wooden Star Christmas Scandinavian” also contain “Wooden Heart Christmas Scandinavian”.
6(iii)
With a lift of 44.05, we can tell that customers who purchase “Wooden Star Christmas Scandinavian” are 44 times more likely to also purchase “Wooden Heart Christmas Scandinavian”. This tells us these two products have a very strong relationship.
6(c)
The rule “{ALARM CLOCK BAKELIKE GREEN} => {ALARM CLOCK BAKELIKE RED}“ is trivial as it states that consumers that purchase a green alarm clock are also likely to purchase a red one. This does not give us much insight as it is the same product in a different color variation. While the type of product tells us that once customers purchase one version of the product they’re not likely to need a second. This is creating a rule based off of the idea that the consumer required the alarm clock product. However, this rule is telling us they’re likely to purchase two. Although this is not the common case.
Another example of a similarly trivial rule is “{LUNCH BAG RED RETROSPOT} => {LUNCH BAG PINK POLKADOT}“. This is trivial for similar reasons to the above. The rule does not seem to give us anything actionable as it is stating that customers will purchase two variations of the same product. If this was a food item situation this could make sense as customers may consider purchasing two different types of berries for example. As they’re both berries the rule would group them together as likely to be purchased together. However, this rule does not seem to take into account the type of product and how it would be purchased. Consumers would search for one variation of a lunch bag and purchase one lunch bag as they would rarely have a need for two. Decreasing the chances of a customer purchasing two lunch bags within the same transaction.
6(d)
I believe the rule “{GARDENERS KNEELING PAD KEEP CALM} => {GARDENERS KNEELING PAD CUP OF TEA}” is actionable as management could sell these products in bundles to consumers that have a”green thumb”. Targeting household gardeners with a product bundle, a buy one get one half price offer or a pop up showcasing “customers who bought GARDENERS KNEELING PAD KEEP CALM also bought GARDENERS KNEELING PAD CUP OF TEA”. Aiming to get their garden back in order for the Summer.
Another similarly actionable rule is “{WOODEN PICTURE FRAME WHITE FINISH} => {WOODEN FRAME ANTIQUE WHITE}”. These products could be marketed in a home decor renovation bundle. Pushing the specifically white variables of products to consumers. Where these products differ from the non actionable trivial products above is the fact that consumers will have multiple photo frames around their homes. Instead of just one, so offering them variation while still sticking to the same theme of white can boost online sales.
7
print(rose_rules_sorted)set of 10 rules inspect(rose_rules_sorted) lhs rhs support confidence coverage lift count
[1] {PINK REGENCY TEACUP AND SAUCER,
ROSES REGENCY TEACUP AND SAUCER} => {GREEN REGENCY TEACUP AND SAUCER} 0.0156 0.8813559 0.0177 28.708662 156
[2] {GREEN REGENCY TEACUP AND SAUCER,
ROSES REGENCY TEACUP AND SAUCER} => {PINK REGENCY TEACUP AND SAUCER} 0.0156 0.6872247 0.0227 28.397714 156
[3] {GREEN REGENCY TEACUP AND SAUCER,
PINK REGENCY TEACUP AND SAUCER} => {ROSES REGENCY TEACUP AND SAUCER} 0.0156 0.8125000 0.0192 24.399399 156
[4] {REGENCY CAKESTAND 3 TIER,
ROSES REGENCY TEACUP AND SAUCER} => {GREEN REGENCY TEACUP AND SAUCER} 0.0116 0.7483871 0.0155 24.377430 116
[5] {GREEN REGENCY TEACUP AND SAUCER,
REGENCY CAKESTAND 3 TIER} => {ROSES REGENCY TEACUP AND SAUCER} 0.0116 0.7891156 0.0147 23.697167 116
[6] {GREEN REGENCY TEACUP AND SAUCER} => {ROSES REGENCY TEACUP AND SAUCER} 0.0227 0.7394137 0.0307 22.204615 227
[7] {ROSES REGENCY TEACUP AND SAUCER} => {GREEN REGENCY TEACUP AND SAUCER} 0.0227 0.6816817 0.0333 22.204615 227
[8] {PINK REGENCY TEACUP AND SAUCER} => {ROSES REGENCY TEACUP AND SAUCER} 0.0177 0.7314050 0.0242 21.964113 177
[9] {ROSES REGENCY TEACUP AND SAUCER} => {PINK REGENCY TEACUP AND SAUCER} 0.0177 0.5315315 0.0333 21.964113 177
[10] {GREEN REGENCY TEACUP AND SAUCER,
ROSES REGENCY TEACUP AND SAUCER} => {REGENCY CAKESTAND 3 TIER} 0.0116 0.5110132 0.0227 6.585222 116The code above produces all 10 rules involving “ROSES REGENCY TEACUP AND SAUCER”.
If a customer purchases “ROSES REGENCY TEACUP AND SAUCER”, they are most likely to purchase “GREEN REGENCY TEACUP AND SAUCER”, “PINK REGENCY TEACUP AND SAUCER” or “REGENCY CAKESTAND 3 TIER”.
Collaborative Filtering
Number 1
library(tidyverse)
library(recommenderlab)
library(dplyr)
library(ggplot2)
library(tibble)Number 2
steam_ratings <- read_csv("steam_ratings.csv")Warning: One or more parsing issues, call `problems()` on your data frame for details,
e.g.:
dat <- vroom(...)
problems(dat)steam_ratings <- as(steam_ratings, "matrix")
steam_ratings <- as(steam_ratings, "realRatingMatrix")Number 3
3(a)
vector_ratings
0 1 2 3 4 5
3236066 4773 12500 19762 10655 4724 - We can see that 3,236,066 have a rating of 0 meaning that they actually have a missing rating. With a rating of 3 being the most common and a rating of 5 being the least common.
3(b)
Warning: Removed 2 rows containing non-finite values (`stat_bin()`).
- We can see from our graph above that the majority of ratings fall between 2-4.
3(c)
user_game_counts <- rowCounts(steam_ratings)ggplot(data = data.frame(game_count = user_game_counts), aes(x = game_count)) +
geom_histogram(binwidth = 20, fill = "skyblue", color = "black") +
labs(title = "Games Rated per User", x = "Total Games Rated", y = "Frequency")
We can see from the above graph that most users have rated between 1-40 games.
However this histogram is skewed due to some outliers who have rated up to 300 games.
Number 4
In an attempt to tidy the data we’re going to do the following.
Remove Less Popular Games: Exclude games with less than 6 ratings. This ensures we focus on games with enough ratings for meaningful recommendations.
Filter Out Less Active Users: Remove users who have rated fewer than 5 games. Users with limited ratings may not provide sufficient information for accurate recommendations
steam_small <- steam_ratings[rowCounts(steam_ratings) >= 5, ]
steam_small <- steam_small[, colCounts(steam_small) >= 6]
steam_small_matrix <- as(steam_small, "matrix")4(a)
set.seed(101)4(b)
eval_steam <- evaluationScheme(data = steam_small,
method = "split",
train = 0.8,
given = 6,
goodRating = 3)Warning in .local(data, ...): Dropping these users from the evaluation since they have fewer rating than specified in given!
These users are 400, 423, 620, 670, 681, 683, 863, 1363, 1391, 1624, 1649, 1735, 1832, 1912, 1966, 2016, 20544(c)
train <- getData(eval_steam, "train") known <- getData(eval_steam, "known") unknown <- getData(eval_steam, "unknown")
Number 5
ubcf_model <- Recommender(data = train,
method = "UBCF",
parameter = list(normalize = "center", method = "Cosine"))ubcf_predict <- predict(object = ubcf_model,
newdata = known,
type = "ratings")ubcf_eval <- calcPredictionAccuracy(x = ubcf_predict,
data = unknown)ubcf_eval RMSE MSE MAE
1.1887736 1.4131826 0.9272125 The MAE is 0.9145851
- This tells us that on average, the predicted ratings are off by approximately 0.91 of a rating.
Number 6
ibcf_model <- Recommender(data = train,
method = "IBCF",
parameter = list(normalize = "center", method = "Cosine"))ibcf_predict <- predict(object = ibcf_model,
newdata = known,
type = "ratings")ibcf_eval <- calcPredictionAccuracy(x = ibcf_predict,
data = unknown)ibcf_eval RMSE MSE MAE
1.492958 2.228922 1.167412 The MAE is 1.148418.
- This tells us that on average the predicted ratings are off by approximately 1.15 of a rating.
Number 7
We can tell that the UBCF model is more accurate than the IBCF model as the MAE for the UBFC model is 0.91, and the IBCF model’s MAE is 1.14. Making the UBCF model more accurate.
ubcf_recs <- predict(object = ubcf_model,
newdata = known,
type = "topNList",
n = 3)ubcf_recs <- predict(object = ubcf_model,
newdata = known,
type = "topNList",
n = 3)
ubcf_recs_list <- as(ubcf_recs, "list")
for (i in 1:5) {print(paste("Recommendations for user", i))
print(ubcf_recs_list[[i]])}[1] "Recommendations for user 1"
[1] "Hitman Blood Money" "Half-Life 2" "Pid"
[1] "Recommendations for user 2"
[1] "Cry of Fear" "SMITE" "NBA 2K14"
[1] "Recommendations for user 3"
[1] "Hitman Blood Money" "Hazard Ops" "Time Clickers"
[1] "Recommendations for user 4"
[1] "Hitman Blood Money" "Time Clickers" "X-COM UFO Defense"
[1] "Recommendations for user 5"
[1] "Hitman Blood Money" "Time Clickers" "Town of Salem" Recommendations for user 1
God Mode
Haunted Memories
Bulletstorm
Recommendations for user 2
God Mode
Half-Life 2 Deathmatch
Hurtworld
Recommendations for user 3
Scania Truck Driving Simulator
The Ultimate DOOM
Gotham City Impostors Free To Play
Recommendations for user 4
Brothers - A Tale of Two Sons
Defense Grid 2
God Mode
Recommendations for user 5
Brothers - A Tale of Two Sons
Grimoire Manastorm
RPG MO
Number 8
Steam could use the UBCF collaborative filtering model in a number of ways.
The most obvious way is to suggest these games to the relative users as “Suggestions for you” in order to increase sales.
Steam could offer users 3 day free trials to users of the recommended games. Giving the user the opportunity to use the game and get “hooked” to a point where at the end of their free trial they’re encouraged to purchase the game.
Another way is to bundle the recommended games into “Sale Packages” that will be advertised to the users in a campaign. With the idea of the campaign being “Buy one get one half price”