Dataset comes from UC Irvine Machine Learning Repository. It contains the data about shopping baskets of customers from different countries around the world. In this excercise the data will be analyzed to find recommendations for customers in future shopping.
data_raw <- read_excel('Data/online_retail_II.xlsx')
Data:
Invoice - Invoice number. Nominal. A 6-digit integral number uniquely
assigned to each transaction. If this code starts with the letter ‘C’,
it indicates a cancellation
StockCode - Product (item) code. Nominal. A 5-digit integral number
uniquely assigned to each distinct product
Description - Product (item) name. Nominal
Quantity - The quantities of each product (item) per transaction.
Numeric
InvoiceDate - Invoice date and time. Numeric. The day and time when a
transaction was generated
UnitPrice - Unit price. Numeric. Product price per unit in sterling
(£)
CustomerID - Customer number. Nominal. A 5-digit integral number
uniquely assigned to each customer
Country - Country name. Nominal. The name of the country where a
customer resides
head(data_raw)
## # A tibble: 6 × 8
## Invoice StockCode Description Quantity InvoiceDate Price `Customer ID`
## <chr> <chr> <chr> <dbl> <dttm> <dbl> <dbl>
## 1 489434 85048 "15CM CHRI… 12 2009-12-01 07:45:00 6.95 13085
## 2 489434 79323P "PINK CHER… 12 2009-12-01 07:45:00 6.75 13085
## 3 489434 79323W "WHITE CHE… 12 2009-12-01 07:45:00 6.75 13085
## 4 489434 22041 "RECORD FR… 48 2009-12-01 07:45:00 2.1 13085
## 5 489434 21232 "STRAWBERR… 24 2009-12-01 07:45:00 1.25 13085
## 6 489434 22064 "PINK DOUG… 24 2009-12-01 07:45:00 1.65 13085
## # ℹ 1 more variable: Country <chr>
data_to_transform <- data_raw
# Convert StockCode column to factor
data_to_transform$StockCode <- as.factor(data_to_transform$StockCode)
# Create a new column with value 1 indicating the existence of the
# combination of Invoice and StockCode
data_to_transform$Exist <- 1
# Check how many unique stock codes there are
length <- length(unique(as.character(data_to_transform$StockCode)))
# 4631 - too many columns
# Display all possible countries
print(unique(data_to_transform$Country))
## [1] "United Kingdom" "France" "USA"
## [4] "Belgium" "Australia" "EIRE"
## [7] "Germany" "Portugal" "Japan"
## [10] "Denmark" "Nigeria" "Netherlands"
## [13] "Poland" "Spain" "Channel Islands"
## [16] "Italy" "Cyprus" "Greece"
## [19] "Norway" "Austria" "Sweden"
## [22] "United Arab Emirates" "Finland" "Switzerland"
## [25] "Unspecified" "Malta" "Bahrain"
## [28] "RSA" "Bermuda" "Hong Kong"
## [31] "Singapore" "Thailand" "Israel"
## [34] "Lithuania" "West Indies" "Lebanon"
## [37] "Korea" "Brazil" "Canada"
## [40] "Iceland"
# Count products per invoice in each country
products_per_invoice <- data_raw %>%
group_by(Country, Invoice) %>%
summarise(total_count=n(),
.groups = 'drop')
# Calculate average products per invoice in each country
products_per_invoice_avg <- products_per_invoice %>%
group_by(Country) %>%
summarise_at(vars(total_count), list(avg_prod = mean))
# Displaying countries with the highest average numbers of products per invoice
products_per_invoice_avg[order(products_per_invoice_avg$avg_prod,
decreasing = T), ]
## # A tibble: 40 × 2
## Country avg_prod
## <chr> <dbl>
## 1 Brazil 62
## 2 West Indies 54
## 3 Canada 38.5
## 4 Thailand 38
## 5 Israel 37
## 6 Iceland 35.5
## 7 Greece 34.5
## 8 Bermuda 34
## 9 Singapore 29.2
## 10 Malta 28.7
## # ℹ 30 more rows
# Count invoices per country
invoices_per_country <- data_raw %>%
group_by(Country) %>%
summarise(distinct_invoices = n_distinct(Invoice))
# Displaying countries with the highest numbers of invoices
invoices_per_country[order(invoices_per_country$distinct_invoices,
decreasing = T), ]
## # A tibble: 40 × 2
## Country distinct_invoices
## <chr> <int>
## 1 United Kingdom 26633
## 2 Germany 513
## 3 EIRE 457
## 4 France 297
## 5 Netherlands 150
## 6 Spain 85
## 7 Sweden 83
## 8 Belgium 65
## 9 Portugal 57
## 10 Australia 50
## # ℹ 30 more rows
# Count StockCodes per country
stock_per_country <- data_raw %>%
group_by(Country) %>%
summarise(distinct_codes = n_distinct(StockCode))
# Displaying countries with the lowest numbers of stock codes
stock_per_country[order(stock_per_country$distinct_codes,
decreasing = F), ]
## # A tibble: 40 × 2
## Country distinct_codes
## <chr> <int>
## 1 Lebanon 13
## 2 Nigeria 32
## 3 Bermuda 34
## 4 West Indies 49
## 5 Korea 52
## 6 Hong Kong 60
## 7 Brazil 61
## 8 Iceland 70
## 9 Israel 73
## 10 Thailand 76
## # ℹ 30 more rows
Germany seem to be country with sufficient number of observations.
data_ger <- data_to_transform[data_to_transform$Country == 'Germany', ]
# Check how many StockCodes there are in Germany
unique(as.character(data_ger[['StockCode']]))
## [1] "85049E" "21976" "21498" "22077" "84946" "84948" "21537"
## [8] "21733" "85123A" "20718" "21539" "16169F" "84836" "21527"
## [15] "21533" "21531" "22180" "22333" "22319" "85232B" "20676"
## [22] "20914" "21217" "21238" "21242" "22232" "21535" "21844"
## [29] "22073" "22072" "22321" "15056BL" "15056N" "21731" "22254"
## [36] "22255" "46000S" "20964" "21788" "21789" "20682" "85014C"
## [43] "84832" "POST" "22198" "85232A" "21491" "21432" "21672"
## [50] "21671" "21658" "22253" "22348" "21559" "21121" "21212"
## [57] "22139" "22352" "22114" "22353" "21576" "21577" "21578"
## [64] "21429" "21481" "21485" "21700" "21973" "20719" "21232"
## [71] "21231" "22313" "22314" "22200" "22301" "22135" "22331"
## [78] "21697" "22138" "21080" "22202" "22150" "21210" "21039"
## [85] "21674" "85038" "22178" "22189" "21693" "22278" "21500"
## [92] "51014L" "22299" "20692" "20702" "20703" "20696" "21284"
## [99] "21286" "21843" "21916" "22354" "20665" "20750" "20681"
## [106] "85049A" "22192" "22233" "21258" "84997A" "84247L" "85014B"
## [113] "20679" "22034" "22042" "22044" "16169C" "21206" "22090"
## [120] "85199L" "85199S" "22219" "22179" "21913" "21556" "21542"
## [127] "16169K" "16169N" "16169P" "22045" "22024" "22025" "22027"
## [134] "22029" "22037" "22075" "22076" "22078" "85049B" "85049F"
## [141] "85049C" "22242" "22243" "22244" "22245" "21241" "21245"
## [148] "20674" "84978" "21669" "21223" "22212" "21482" "21945"
## [155] "22355" "22302" "22303" "21993" "21580" "21124" "84247C"
## [162] "84247K" "21892" "22028" "22064" "37495" "22318" "22131"
## [169] "22344" "21944" "21785" "21027" "21842" "85099F" "21928"
## [176] "21929" "21051" "21249" "21154" "21216" "85099B" "84997D"
## [183] "84997C" "84997B" "20754" "21673" "21668" "21285" "21675"
## [190] "84692" "21077" "22332" "21380" "47590A" "21126" "21125"
## [197] "20713" "20711" "21440" "84568" "21680" "21918" "21791"
## [204] "21525" "22203" "22199" "84029G" "22193" "21717" "21164"
## [211] "22349" "22350" "85174" "21115" "22283" "21980" "90167"
## [218] "20675" "20677" "22236" "22231" "84464" "35922" "22252"
## [225] "84879" "21157" "20871" "20726" "21276" "21275" "20728"
## [232] "79191C" "21936" "20981" "21426" "21424" "21431" "84569C"
## [239] "22276" "21935" "22277" "84592" "20724" "20723" "21931"
## [246] "47594A" "20658" "20684" "21484" "22320" "85206A" "21439"
## [253] "22181" "21156" "47592A" "47591D" "20712" "21383" "22021"
## [260] "84437A" "85205B" "85206B" "22326" "22328" "22329" "22343"
## [267] "81952V" "22082" "21937" "85059" "35809A" "35811A" "84968A"
## [274] "81950B" "79072" "85162C" "84968B" "47559B" "85162B" "84279B"
## [281] "84279P" "35241" "81952B" "20979" "21035" "21752" "21667"
## [288] "85093" "84531B" "84691" "21155" "22218" "22215" "21670"
## [295] "20652" "20653" "21706" "47367B" "21652" "22083" "22230"
## [302] "22251" "84466" "21496" "35970" "47594B" "22358" "21240"
## [309] "21239" "21243" "21244" "84270" "84968D" "84968C" "84968F"
## [316] "84969" "20685" "20717" "20725" "20983" "21041" "21058"
## [323] "21059" "21078" "21085" "21086" "21091" "21092" "21094"
## [330] "21122" "21153" "21218" "21329" "21330" "21541" "21655"
## [337] "21768" "21880" "21894" "21896" "21972" "21975" "22079"
## [344] "22334" "22356" "84970S" "85015" "85016" "85032A" "85032B"
## [351] "85055" "47590B" "20857" "20864" "21226" "16237" "20751"
## [358] "21563" "M" "21394" "21395" "21392" "21393" "85221"
## [365] "22367" "21381" "21130" "47471" "20738" "21592" "21588"
## [372] "22197" "ADJUST" "22201" "21881" "21884" "46126A" "84854"
## [379] "85216" "85217" "21144" "47570B" "47587A" "84510B" "22133"
## [386] "79320" "20716" "20729" "20730" "20966" "20963" "21942"
## [393] "21943" "21946" "21165" "22117" "84676" "22204" "21042"
## [400] "21246" "20672" "22058" "22059" "22062" "22063" "37448"
## [407] "37450" "47556B" "21754" "22127" "22273" "21746" "21749"
## [414] "16256B" "20977" "22080" "16052" "21204" "85049H" "20990"
## [421] "21949" "35810A" "48194" "21837" "22031" "22033" "22040"
## [428] "22038" "21610" "21612" "84932D" "84932C" "84932A" "84932B"
## [435] "84199" "21401" "22500" "84937" "21445" "21446" "21447"
## [442] "21529" "22366" "22497" "22495" "22071" "22070" "82581"
## [449] "84927A" "85184D" "85007" "21414" "21415" "21416" "22325"
## [456] "20734" "22417" "21428" "85231B" "85231E" "85231L" "21763"
## [463] "21891" "21709" "21708" "21561" "21555" "48185" "84580"
## [470] "40046A" "21100" "22385" "21930" "84228" "84461" "84218"
## [477] "85215" "10002" "84012" "22377" "22378" "22379" "22381"
## [484] "22384" "22386" "20972" "22149" "22274" "22272" "22271"
## [491] "84625C" "15060B" "47585A" "84991" "22055" "20828" "21099"
## [498] "21499" "16169M" "85049D" "22128" "22129" "22382" "21883"
## [505] "22322" "21534" "22260" "21397" "21396" "21528" "22107"
## [512] "84564" "22300" "21398" "21399" "21906" "71403" "84670"
## [519] "21064" "84575B" "21579" "22264" "22256" "84247G" "22032"
## [526] "22411" "85226A" "85136A" "22431" "21219" "22432" "22371"
## [533] "22374" "22375" "22376" "21715" "35910A" "84219" "85200"
## [540] "21648" "84905" "84909A" "84657" "47586A" "84467" "84469"
## [547] "21524" "47570" "16161P" "20758" "20772" "20768" "20767"
## [554] "20755" "20978" "20975" "85175" "21889" "37370" "84934"
## [561] "21637" "20759" "20769" "20770" "22433" "22136" "21213"
## [568] "22327" "47579" "22067" "84509G" "84509B" "37333" "37500"
## [575] "21211" "84987" "37447" "22134" "47502" "20689" "37501"
## [582] "37464" "82613A" "82613D" "21137" "84709B" "21278" "37482P"
## [589] "21873" "21872" "84597C" "22094" "84534B" "21720" "21721"
## [596] "21722" "21725" "16156S" "16161G" "21495" "22046" "84533B"
## [603] "84522" "37503" "20661" "20799" "20798" "21207" "21374"
## [610] "22383" "20749" "47505" "21082" "22467" "22466" "22468"
## [617] "22609" "21723" "22509" "22505" "22478" "22477" "22472"
## [624] "22457" "84988" "21653" "84970L" "21679" "21830" "15036"
## [631] "40016" "85227" "21640" "21716" "84529C" "47591B" "22507"
## [638] "22479" "47504J" "21411" "22315" "21127" "21724" "16238"
## [645] "20971" "22568" "22569" "22570" "22450" "22451" "22449"
## [652] "20619" "22473" "22475" "22471" "20998" "22228" "22229"
## [659] "85169C" "85169D" "21319" "21320" "21661" "21662" "22423"
## [666] "22585" "21981" "16168M" "22476" "22583" "22584" "22445"
## [673] "22560" "22556" "22555" "48189" "48195" "84815" "22446"
## [680] "84536A" "21471" "21461" "21348" "85049G" "21564" "21452"
## [687] "21110" "21281" "22279" "20866" "82486" "22508" "22511"
## [694] "85129A" "22317" "85164B" "20794" "21272" "35810B" "84050"
## [701] "20771" "22111" "21523" "48116" "48129" "48187" "48188"
## [708] "48173C" "48184" "85162A" "22122" "22389" "22521" "22522"
## [715] "22523" "22444" "22559" "22561" "21718" "22540" "22544"
## [722] "22551" "22554" "22549" "22515" "22516" "22517" "22518"
## [729] "22519" "22520" "22454" "22453" "22531" "22534" "21914"
## [736] "21915" "84715" "22036" "21507" "21251" "85232D" "22557"
## [743] "21002" "21001" "16161U" "21621" "21162" "21114" "22403"
## [750] "22398" "22402" "21696" "22480" "22482" "21488" "22309"
## [757] "22582" "21690" "85040B" "21222" "21220" "22536" "22532"
## [764] "17091A" "21866" "21202" "21203" "84247N" "16236" "10125"
## [771] "78124" "22530" "22562" "22563" "22564" "22567" "22399"
## [778] "22405" "22617" "22618" "47589" "21614" "22447" "22452"
## [785] "21983" "21984" "21136" "21932" "21990" "21992" "22336"
## [792] "21786" "21787" "22611" "22023" "21509" "22047" "21922"
## [799] "21925" "48138" "21586" "21590" "21877" "21871" "21069"
## [806] "21067" "21609" "21620" "21615" "PADS" "46000M" "22406"
## [813] "22409" "22360" "22359" "21875" "22612" "21878" "22084"
## [820] "22357" "22499" "21622" "21623" "22494" "22442" "22443"
## [827] "22316" "21987" "21988" "21989" "21224" "21703" "15039"
## [834] "22169" "72741" "22393" "22390" "22448" "35600D" "35598D"
## [841] "84341B" "21977" "22416" "21974" "21829" "79029" "22606"
## [848] "22605" "22400" "22410" "22496" "84029E" "47566" "21741"
## [855] "22345" "22194" "21354" "40018F" "22081" "84797B" "84797A"
## [862] "21197" "20961" "22436" "22439" "22211" "22526" "22525"
## [869] "22524" "22487" "22460" "85017A" "21135" "84581" "21735"
## [876] "79030G" "21307" "21474" "22069" "21833" "22627" "22625"
## [883] "22504" "22503" "22489" "22491" "22626" "21719" "21562"
## [890] "84899E" "15044B" "15044D" "22498" "22088" "47566B" "21096"
## [897] "21088" "79321" "20973" "22275" "22624" "85099C" "21882"
## [904] "21327" "84828" "46118" "85119" "21117" "22492" "22621"
## [911] "22622" "15056P" "35015" "22420" "84949" "85062" "22514"
## [918] "15058A" "21714" "17109D" "72750B" "47567B" "85090" "22427"
## [925] "22493" "22424" "22429" "22426" "22425" "22430" "84086C"
## [932] "10135" "16045" "22634" "22635" "22607" "21832" "84675"
## [939] "21923" "22089" "22533" "21328" "21986" "21933" "20974"
## [946] "20752" "84598" "22440" "22435" "22613" "22649" "22636"
## [953] "22646" "22644" "21558" "84546" "47421" "22502" "22501"
## [960] "22324" "22704" "22705" "22711" "22716" "21508" "22437"
## [967] "22614" "22615" "22620" "21205" "21704" "47591A" "85178"
## [974] "22637" "21033" "85132C" "20985" "21503" "85098B" "84212"
## [981] "22419" "22068" "84520B" "72140F" "72351A" "72351B" "21215"
## [988] "84989A" "21291" "47504K" "79030A" "22558" "84840" "21742"
## [995] "72760B" "21982" "22418" "37474" "21108" "21190" "82578"
## [1002] "84380" "47568" "15034" "84839" "85068" "35637A" "44230"
## [1009] "84356" "20829" "21836" "84968E" "22262" "84509A" "84913A"
## [1016] "84347" "84247E" "35833G" "22086" "35964" "21494" "35400"
## [1023] "35597B" "35598B" "35599B" "22365" "22414" "48197" "22465"
## [1030] "20654" "20727" "21084" "21947" "21967" "22035" "22171"
## [1037] "22650" "72728" "72819" "84929" "85006" "85017B" "85017C"
## [1044] "85031A" "85031B" "85054" "21090" "21260" "21358" "21506"
## [1051] "22553" "22628" "84912A" "84819" "84818" "22630" "21410"
## [1058] "79030D" "85159B" "21775" "84755" "21687" "21691" "85064"
## [1065] "85067" "21625" "85121C" "21259" "20997" "84658" "21544"
## [1072] "22052" "21519" "21581" "22546" "22529" "22535" "22537"
## [1079] "22547" "85226C" "22545" "22629" "22631" "22669" "22596"
## [1086] "22602" "22600" "22603" "22595" "22593" "22130" "21802"
## [1093] "22588" "22589" "22591" "21188" "71459" "84992" "22030"
## [1100] "21497" "22438" "85132A" "85132B" "79190B" "79190D" "21221"
## [1107] "22667" "22665" "22666" "22488" "21917" "22619" "21106"
## [1114] "22539" "22364" "20886" "21282" "21611" "79000" "21324"
## [1121] "22485" "21827" "22616" "21890" "21845" "84792" "82494L"
## [1128] "21361" "82484" "22312" "84849B" "84847" "21116" "84006"
## [1135] "22652" "22659" "22668" "20982" "11001" "22176" "22571"
## [1142] "22581" "22578" "22752" "22757" "22753" "22754" "84859A"
## [1149] "22759" "22743" "22654" "21430" "22718" "22742" "22744"
## [1156] "84899F" "51008" "47593A" "22661" "22712" "22483" "22692"
## [1163] "84661C" "84661B" "84661A" "22380" "22601" "22116" "22413"
## [1170] "21163" "21248" "22391" "21357" "21355" "22132" "22108"
## [1177] "22124" "21025" "21026" "21705" "22221" "21011" "22155"
## [1184] "22653" "21263" "82616C" "22702" "22700" "22697" "22698"
## [1191] "21269" "21270" "22691" "21707" "84510A" "21651" "20984"
## [1198] "85087" "35951" "21000" "84912B" "84685" "82580" "35832"
## [1205] "84945" "47422" "84507C" "22835" "22865" "22867" "37340"
## [1212] "22813" "22820" "16012" "22749" "22951" "22949" "22908"
## [1219] "22907" "22894" "22897" "22806" "22816" "22809" "22812"
## [1226] "22945" "22734" "22739" "22952" "22950" "22909" "22733"
## [1233] "22731" "21056" "22464" "21739" "22456" "22725" "22726"
## [1240] "22727" "22730" "22191" "21247" "22170" "22719" "22895"
## [1247] "22898" "22814" "22815" "22821" "21486" "84030E" "22074"
## [1254] "22207" "79191B" "21070" "21464" "22756" "22834" "70007"
## [1261] "22939" "22113" "21098" "21645" "22857" "22860" "22859"
## [1268] "22538" "84031A" "22910" "90200E" "20671" "22807" "47518F"
## [1275] "22755" "82613B" "22323" "62094B" "21166" "21903" "22746"
## [1282] "22745" "22747" "22748" "22741" "22728" "22938" "22572"
## [1289] "21186" "22346" "47599B" "16259" "84077" "16033" "84826"
## [1296] "47599A" "22896" "22900" "35599D" "21618" "21619" "84378"
## [1303] "22707" "82482" "22294" "22173" "22714" "22819" "22984"
## [1310] "35095A" "21790" "35471D" "35471A" "85032D" "72816" "22946"
## [1317] "22651" "22758" "22043" "22959" "22295" "22304" "21191"
## [1324] "84947" "22766" "22632" "22764" "22849" "22847" "22788"
## [1331] "22740" "22843" "22842" "22904" "22750" "22732" "22940"
## [1338] "22168" "22892" "82551" "22717" "22671" "22675" "84508B"
## [1345] "22175" "84078A" "20816" "22735" "22737" "22594" "22145"
## [1352] "22706" "15044A" "22837" "22049" "22099" "22699" "22781"
## [1359] "22799" "71053" "22690" "22845" "22844" "22840" "22841"
## [1366] "22968" "22110" "20764" "20766" "20760" "20762" "20756"
## [1373] "20761" "20622" "20617" "84406B" "21467" "21472" "22846"
## [1380] "22396" "21870" "22891" "17003" "90018A" "90019A" "21755"
## [1387] "22664" "22663" "22902" "22801" "22798" "22818" "22736"
## [1394] "22738" "22573" "22574" "22576" "22577" "22579" "21801"
## [1401] "21803" "22687" "22580" "20957" "22779" "84569B" "22307"
## [1408] "71477" "21730" "22751" "22941" "22890" "22776" "21179"
## [1415] "84429A" "22729" "22137" "22633" "22953" "22906" "22810"
## [1422] "22852" "22853" "22141" "21479" "90104" "22929" "22964"
## [1429] "22966" "20967" "22956" "20867" "20868" "22794" "22913"
## [1436] "22923" "22969" "22962" "22961" "21756" "22922" "22924"
## [1443] "22973" "22767" "22963" "22926" "22914" "84849D" "16016"
## [1450] "22960" "22972" "22975" "22976" "84375" "20668" "21817"
## [1457] "22337" "22144" "84920" "22943" "22550"
# 1400 ok
# Check numbers of each stockcode in Germany
num_of_stocks <- data_ger %>%
group_by(StockCode) %>%
summarise(distinct_invoices = n_distinct(Invoice))
# Display products with high number of purchases
num_of_stocks[order(num_of_stocks$distinct_invoices, decreasing = T), ]
## # A tibble: 1,461 × 2
## StockCode distinct_invoices
## <fct> <int>
## 1 POST 280
## 2 22326 103
## 3 20719 77
## 4 22328 69
## 5 21731 66
## 6 20914 58
## 7 21238 57
## 8 22243 52
## 9 22242 51
## 10 21578 50
## # ℹ 1,451 more rows
# I'll select products that appear at least 50 times in the data
products <- num_of_stocks[num_of_stocks[['distinct_invoices']] >= 50 &
num_of_stocks[['StockCode']] != 'POST', ]
# Filtering the data to only most popular products
data_filtered <- data_ger[data_ger[['StockCode']] %in% products[['StockCode']],
c('Invoice', 'Description')]
for(i in colnames(data_filtered)) {
data_filtered[[i]] = as.factor(data_filtered[[i]])
}
# Transforming the data to transaction
transactions <- transactions(data_filtered, format = 'long')
summary(transactions)
## transactions as itemMatrix in sparse format with
## 242 rows (elements/itemsets/transactions) and
## 15 columns (items) and a density of 0.1606061
##
## most frequent items:
## ROUND SNACK BOXES SET OF4 WOODLAND WOODLAND CHARLOTTE BAG
## 102 77
## ROUND SNACK BOXES SET OF 4 FRUITS RED TOADSTOOL LED NIGHT LIGHT
## 67 66
## WOODLAND DESIGN COTTON TOTE BAG (Other)
## 50 221
##
## element (itemset/transaction) length distribution:
## sizes
## 1 2 3 4 5 6
## 79 65 52 23 12 11
##
## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 1.000 1.000 2.000 2.409 3.000 6.000
##
## includes extended item information - examples:
## labels
## 1 5 HOOK HANGER MAGIC TOADSTOOL
## 2 5 HOOK HANGER RED MAGIC TOADSTOOL
## 3 HOOK, 5 HANGER , MAGIC TOADSTOOL
##
## includes extended transaction information - examples:
## transactionID
## 1 489526
## 2 490395
## 3 490563
# Displaying simple statistics
itemFrequency(transactions, type="relative")
## 5 HOOK HANGER MAGIC TOADSTOOL 5 HOOK HANGER RED MAGIC TOADSTOOL
## 0.061983471 0.078512397
## HOOK, 5 HANGER , MAGIC TOADSTOOL HOOK, 5 HANGER ,MAGIC TOADSTOOL RED
## 0.148760331 0.136363636
## RED RETROSPOT CUP RED SPOTTY CUP
## 0.041322314 0.194214876
## RED TOADSTOOL LED NIGHT LIGHT ROUND SNACK BOXES ,SET 4, FRUITS
## 0.272727273 0.008264463
## ROUND SNACK BOXES ,SET4, WOODLAND ROUND SNACK BOXES SET OF 4 FRUITS
## 0.004132231 0.276859504
## ROUND SNACK BOXES SET OF4 WOODLAND SET/5 RED RETROSPOT LID GLASS BOWLS
## 0.421487603 0.053719008
## SET/5 RED SPOTTY LID GLASS BOWLS WOODLAND CHARLOTTE BAG
## 0.185950413 0.318181818
## WOODLAND DESIGN COTTON TOTE BAG
## 0.206611570
itemFrequency(transactions, type="absolute")
## 5 HOOK HANGER MAGIC TOADSTOOL 5 HOOK HANGER RED MAGIC TOADSTOOL
## 15 19
## HOOK, 5 HANGER , MAGIC TOADSTOOL HOOK, 5 HANGER ,MAGIC TOADSTOOL RED
## 36 33
## RED RETROSPOT CUP RED SPOTTY CUP
## 10 47
## RED TOADSTOOL LED NIGHT LIGHT ROUND SNACK BOXES ,SET 4, FRUITS
## 66 2
## ROUND SNACK BOXES ,SET4, WOODLAND ROUND SNACK BOXES SET OF 4 FRUITS
## 1 67
## ROUND SNACK BOXES SET OF4 WOODLAND SET/5 RED RETROSPOT LID GLASS BOWLS
## 102 13
## SET/5 RED SPOTTY LID GLASS BOWLS WOODLAND CHARLOTTE BAG
## 45 77
## WOODLAND DESIGN COTTON TOTE BAG
## 50
# Check item frequency
itemFrequency(transactions)
## 5 HOOK HANGER MAGIC TOADSTOOL 5 HOOK HANGER RED MAGIC TOADSTOOL
## 0.061983471 0.078512397
## HOOK, 5 HANGER , MAGIC TOADSTOOL HOOK, 5 HANGER ,MAGIC TOADSTOOL RED
## 0.148760331 0.136363636
## RED RETROSPOT CUP RED SPOTTY CUP
## 0.041322314 0.194214876
## RED TOADSTOOL LED NIGHT LIGHT ROUND SNACK BOXES ,SET 4, FRUITS
## 0.272727273 0.008264463
## ROUND SNACK BOXES ,SET4, WOODLAND ROUND SNACK BOXES SET OF 4 FRUITS
## 0.004132231 0.276859504
## ROUND SNACK BOXES SET OF4 WOODLAND SET/5 RED RETROSPOT LID GLASS BOWLS
## 0.421487603 0.053719008
## SET/5 RED SPOTTY LID GLASS BOWLS WOODLAND CHARLOTTE BAG
## 0.185950413 0.318181818
## WOODLAND DESIGN COTTON TOTE BAG
## 0.206611570
# Plot frequency
itemFrequencyPlot(transactions, support = 0.1)
# Plot top frequency items
itemFrequencyPlot(transactions, topN = 15)
# Calculate AR
ar <- apriori(transactions,
parameter = list(support = 0.006,
confidence = 0.25,
minlen = 2))
## Apriori
##
## Parameter specification:
## confidence minval smax arem aval originalSupport maxtime support minlen
## 0.25 0.1 1 none FALSE TRUE 5 0.006 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: 1
##
## set item appearances ...[0 item(s)] done [0.00s].
## set transactions ...[15 item(s), 242 transaction(s)] done [0.00s].
## sorting and recoding items ... [14 item(s)] done [0.00s].
## creating transaction tree ... done [0.00s].
## checking subsets of size 1 2 3 4 5 done [0.00s].
## writing ... [572 rule(s)] done [0.00s].
## creating S4 object ... done [0.00s].
summary(ar)
## set of 572 rules
##
## rule length distribution (lhs + rhs):sizes
## 2 3 4 5
## 52 225 245 50
##
## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 2.000 3.000 4.000 3.512 4.000 5.000
##
## summary of quality measures:
## support confidence coverage lift
## Min. :0.008264 Min. :0.2500 Min. :0.008264 Min. : 0.7857
## 1st Qu.:0.008264 1st Qu.:0.3383 1st Qu.:0.016529 1st Qu.: 1.4235
## Median :0.012397 Median :0.5000 Median :0.024793 Median : 1.8857
## Mean :0.019426 Mean :0.5365 Mean :0.045917 Mean : 2.7151
## 3rd Qu.:0.020661 3rd Qu.:0.6667 3rd Qu.:0.046488 3rd Qu.: 2.7500
## Max. :0.210744 Max. :1.0000 Max. :0.421488 Max. :16.1333
## count
## Min. : 2.000
## 1st Qu.: 2.000
## Median : 3.000
## Mean : 4.701
## 3rd Qu.: 5.000
## Max. :51.000
##
## mining info:
## data ntransactions support confidence
## transactions 242 0.006 0.25
## call
## apriori(data = transactions, parameter = list(support = 0.006, confidence = 0.25, minlen = 2))
# Inspect the rules
inspect(ar[1:10])
## lhs rhs support confidence coverage lift count
## [1] {RED RETROSPOT CUP} => {5 HOOK HANGER RED MAGIC TOADSTOOL} 0.02066116 0.5000000 0.04132231 6.3684211 5
## [2] {5 HOOK HANGER RED MAGIC TOADSTOOL} => {RED RETROSPOT CUP} 0.02066116 0.2631579 0.07851240 6.3684211 5
## [3] {RED RETROSPOT CUP} => {5 HOOK HANGER MAGIC TOADSTOOL} 0.02066116 0.5000000 0.04132231 8.0666667 5
## [4] {5 HOOK HANGER MAGIC TOADSTOOL} => {RED RETROSPOT CUP} 0.02066116 0.3333333 0.06198347 8.0666667 5
## [5] {RED RETROSPOT CUP} => {WOODLAND CHARLOTTE BAG} 0.01652893 0.4000000 0.04132231 1.2571429 4
## [6] {RED RETROSPOT CUP} => {ROUND SNACK BOXES SET OF4 WOODLAND} 0.02066116 0.5000000 0.04132231 1.1862745 5
## [7] {SET/5 RED RETROSPOT LID GLASS BOWLS} => {WOODLAND DESIGN COTTON TOTE BAG} 0.01652893 0.3076923 0.05371901 1.4892308 4
## [8] {SET/5 RED RETROSPOT LID GLASS BOWLS} => {WOODLAND CHARLOTTE BAG} 0.02479339 0.4615385 0.05371901 1.4505495 6
## [9] {SET/5 RED RETROSPOT LID GLASS BOWLS} => {ROUND SNACK BOXES SET OF4 WOODLAND} 0.02066116 0.3846154 0.05371901 0.9125189 5
## [10] {5 HOOK HANGER RED MAGIC TOADSTOOL} => {5 HOOK HANGER MAGIC TOADSTOOL} 0.02892562 0.3684211 0.07851240 5.9438596 7
# Order them by lift
inspect(sort(ar, by = "lift")[1:5])
## lhs rhs support confidence coverage lift count
## [1] {5 HOOK HANGER RED MAGIC TOADSTOOL,
## WOODLAND CHARLOTTE BAG} => {RED RETROSPOT CUP} 0.008264463 0.6666667 0.012396694 16.13333 2
## [2] {5 HOOK HANGER RED MAGIC TOADSTOOL,
## WOODLAND CHARLOTTE BAG} => {5 HOOK HANGER MAGIC TOADSTOOL} 0.012396694 1.0000000 0.012396694 16.13333 3
## [3] {5 HOOK HANGER RED MAGIC TOADSTOOL,
## RED RETROSPOT CUP,
## WOODLAND CHARLOTTE BAG} => {5 HOOK HANGER MAGIC TOADSTOOL} 0.008264463 1.0000000 0.008264463 16.13333 2
## [4] {5 HOOK HANGER MAGIC TOADSTOOL,
## 5 HOOK HANGER RED MAGIC TOADSTOOL,
## WOODLAND CHARLOTTE BAG} => {RED RETROSPOT CUP} 0.008264463 0.6666667 0.012396694 16.13333 2
## [5] {ROUND SNACK BOXES SET OF 4 FRUITS,
## SET/5 RED RETROSPOT LID GLASS BOWLS,
## WOODLAND CHARLOTTE BAG} => {5 HOOK HANGER MAGIC TOADSTOOL} 0.008264463 1.0000000 0.008264463 16.13333 2
For the sake of plotting selecting less products is better (more
visible), for recommendations let’s take more products.
I’ll select products that appear at least 10 time in the data.
products <- num_of_stocks[num_of_stocks[['distinct_invoices']] >= 10 &
num_of_stocks[['StockCode']] != 'POST', ]
# Filtering the data to only 25 most popular products
data_filtered <- data_ger[data_ger[['StockCode']] %in% products[['StockCode']],
c('Invoice', 'Description')]
for(i in colnames(data_filtered)) {
data_filtered[[i]] = as.factor(data_filtered[[i]])
}
# Transforming the data to transaction
transactions <- transactions(data_filtered, format = 'long')
# Calculate AR
ar <- apriori(transactions,
parameter = list(support = 0.006,
confidence = 0.25,
minlen = 2,
maxlen = 100,
maxtime = 20))
## Apriori
##
## Parameter specification:
## confidence minval smax arem aval originalSupport maxtime support minlen
## 0.25 0.1 1 none FALSE TRUE 20 0.006 2
## maxlen target ext
## 100 rules TRUE
##
## Algorithmic control:
## filter tree heap memopt load sort verbose
## 0.1 TRUE TRUE FALSE TRUE 2 TRUE
##
## Absolute minimum support count: 2
##
## set item appearances ...[0 item(s)] done [0.00s].
## set transactions ...[308 item(s), 439 transaction(s)] done [0.00s].
## sorting and recoding items ... [278 item(s)] done [0.00s].
## creating transaction tree ... done [0.00s].
## checking subsets of size 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 done [0.05s].
## writing ... [1860292 rule(s)] done [0.10s].
## creating S4 object ... done [0.24s].
summary(ar)
## set of 1860292 rules
##
## rule length distribution (lhs + rhs):sizes
## 2 3 4 5 6 7 8 9 10 11 12
## 4331 51083 129652 220734 300972 336889 311360 238149 149700 75944 30228
## 13 14 15 16
## 9061 1918 255 16
##
## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 2.000 6.000 7.000 7.217 9.000 16.000
##
## summary of quality measures:
## support confidence coverage lift
## Min. :0.006834 Min. :0.2500 Min. :0.006834 Min. : 1.076
## 1st Qu.:0.006834 1st Qu.:1.0000 1st Qu.:0.006834 1st Qu.: 13.303
## Median :0.006834 Median :1.0000 Median :0.006834 Median : 17.560
## Mean :0.007023 Mean :0.9523 Mean :0.007721 Mean : 17.599
## 3rd Qu.:0.006834 3rd Qu.:1.0000 3rd Qu.:0.006834 3rd Qu.: 21.950
## Max. :0.116173 Max. :1.0000 Max. :0.232346 Max. :109.750
## count
## Min. : 3.000
## 1st Qu.: 3.000
## Median : 3.000
## Mean : 3.083
## 3rd Qu.: 3.000
## Max. :51.000
##
## mining info:
## data ntransactions support confidence
## transactions 439 0.006 0.25
## call
## apriori(data = transactions, parameter = list(support = 0.006, confidence = 0.25, minlen = 2, maxlen = 100, maxtime = 20))
# Inspect the rules
inspect(subset(ar, count >= 10)[1:10])
## lhs rhs support confidence coverage lift count
## [1] {FELT TOADSTOOL SMALL} => {FELT TOADSTOOL LARGE} 0.02733485 0.9230769 0.02961276 31.171598 12
## [2] {FELT TOADSTOOL LARGE} => {FELT TOADSTOOL SMALL} 0.02733485 0.9230769 0.02961276 31.171598 12
## [3] {SWEETHEART CERAMIC TRINKET BOX} => {STRAWBERRY CERAMIC TRINKET BOX} 0.02733485 0.8000000 0.03416856 10.034286 12
## [4] {STRAWBERRY CERAMIC TRINKET BOX} => {SWEETHEART CERAMIC TRINKET BOX} 0.02733485 0.3428571 0.07972665 10.034286 12
## [5] {TV DINNER TRAY DOLLY GIRL} => {BIG DOUGHNUT FRIDGE MAGNETS} 0.02277904 0.7142857 0.03189066 9.502165 10
## [6] {BIG DOUGHNUT FRIDGE MAGNETS} => {TV DINNER TRAY DOLLY GIRL} 0.02277904 0.3030303 0.07517084 9.502165 10
## [7] {TV DINNER TRAY DOLLY GIRL} => {MEMO BOARD COTTAGE DESIGN} 0.02277904 0.7142857 0.03189066 9.222689 10
## [8] {MEMO BOARD COTTAGE DESIGN} => {TV DINNER TRAY DOLLY GIRL} 0.02277904 0.2941176 0.07744875 9.222689 10
## [9] {GREEN SPOTTY BOWL} => {GREEN SPOTTY CUP} 0.02277904 0.7142857 0.03189066 18.445378 10
## [10] {GREEN SPOTTY CUP} => {GREEN SPOTTY BOWL} 0.02277904 0.5882353 0.03872437 18.445378 10
# Order them by lift
inspect(sort(subset(ar, count >= 10), by = "lift")[1:5])
## lhs rhs support confidence coverage lift count
## [1] {FELT TOADSTOOL SMALL} => {FELT TOADSTOOL LARGE} 0.02733485 0.9230769 0.02961276 31.17160 12
## [2] {FELT TOADSTOOL LARGE} => {FELT TOADSTOOL SMALL} 0.02733485 0.9230769 0.02961276 31.17160 12
## [3] {GREEN SPOTTY CUP,
## RED SPOTTY BOWL} => {GREEN SPOTTY BOWL} 0.02277904 0.9090909 0.02505695 28.50649 10
## [4] {BLUE SPOTTY BOWL,
## PINK SPOTTY PLATE} => {BLUE SPOTTY PLATE} 0.02733485 1.0000000 0.02733485 25.82353 12
## [5] {PINK SPOTTY PLATE,
## RED SPOTTY BOWL} => {BLUE SPOTTY PLATE} 0.02505695 1.0000000 0.02505695 25.82353 11
Recommendations seem reasonable. Most of them are similar items which bring more value if purchased together, because they make a set.