USER MODELING AND USER-ADAPTED INTERACTION - The Journal of Personalization Research
Session-based item recommendation in e-commerce:
on short-term intents, reminders, trends and discounts

Additional Information: Complete List of Examined Features and the Corresponding Feature Weights


Dietmar Jannach
TU Dortmund, Germany
dietmar.jannach [at] tu-dortmund.de
Malte Ludewig
TU Dortmund, Germany
malte.ludewig [at] tu-dortmund.de
Lukas Lerche
TU Dortmund, Germany
lukas.lerche [at] tu-dortmund.de

In our article Session-based item recommendation in e-commerce: on short-term intents, reminders, trends and discounts accepted for publication in User Modeling and User-Adapted Interaction (UMUAI) we examined a classification-based approach to analyze what makes a recommendation successful. In the process we generated over 95 features for each single recommendation action in our data set provided by the online fashion retailer Zalando. Due to space issues we could only explain some of the most relevant features in the article itself. As an addition, the following table lists all investigated features in detail. Furthermore, in our article we reported the top ten feature weights regarding the label prediction calculated by the methods Gain ratio and Chi-squared to highlight the most important success signals. Here, we additionally reveal the weights for all features and also include the Information gain ratio and the Gini index.

Feature Name Type Explanation
clicked Label Recommended item was clicked
clicked_wished Label Recommended item was clicked and added to the wish list
clicked_cart Label Recommended item was clicked and added to the cart
clicked_bought Label Recommended item was clicked and bought in the same session
relpop Numerical Overall popularity of the item
relpop_{day,week,month} Numerical Popularity of the item on the same day, in the same week, or in the same month
samebrand Bool Same brand as the currently viewed item
brandratio_session{1,2,3} Numerical Ratio of the recommended brand regarding actions in the last 1, 2, or 3 sessions
brandpop Numerical Overall popularity of the brand
brandpop_{day,week,month} Numerical Popularity of the brand on the same day, in the same week, or in the same month
samecolor Bool Same color as the currently viewed item
colorratio_session{1,2,3} Numerical Ratio of the recommended color regarding actions in the last 1, 2, or 3 sessions
colorpop Numerical Overall popularity of the color
colorpop_{day,week,month} Numerical Popularity of the color on the same day, in the same week, or in the same month
samecat_{1,2,3,4} Bool Same category as the viewed item on breadcrumb navigation level 1, 2, 3, or 4
catratio_session{1,2,3}_{1,2,3,4} Numerical Ratio of the recommended category (breadcrumb navigation level 1, 2, 3, or 4) regarding actions in the last 1, 2, or 3 sessions
catpop_{1,2,3,4} Numerical Overall popularity of the category on breadcrumb navigation level 1, 2, 3, or 4
catpop_{day,week,month}_{1,2,3,4} Numerical Popularity of the category on the same day, in the same week, or in the same month for breadcrumb navigation level 1, 2, 3, or 4
sameprice Bool Same price level as the currently viewed item
priceratio_session{1,2,3} Numerical Ratio of the recommended price level regarding actions in the last 1, 2, or 3 sessions
similarity_viewed Numerical Ratio of features matched with the currently viewed item
similarity_session{1,2,3} Numerical Average ratio of features matched with items from the last 1, 2, or 3 sessions
neighbors_color Numerical Ratio of neighbor recommendations with the same color
neighbors_brand Numerical Ratio of neighbor recommendations with the same brand
neighbors_price Numerical Ratio of neighbor recommendations with the same price level
neighbors_category_{1,2,3,4} Numerical Ratio of neighbor recommendations with the same category (breadcrumb navigation level 1, 2, 3, or 4)
neighbors_distance Numerical Average ratio of item features matched with neighbor recommendations
prevrecclicks_sim Numerical Average ratio of item features matched with previously clicked recommended items in a session
prevrecclicks_color Numerical Average ratio of matching colors with previously clicked recommended items in a session
prevrecclicks_brand Numerical Average ratio of matching brands with previously clicked recommended items in a session
prevrecclicks_cat_{1,2,3,4} Numerical Average ratio of matching categories with previously clicked recommended items in a session (breadcrumb navigation level 1, 2, 3, or 4)
boughtbefore_sim Numerical Average ratio of item features matched with the last three previously bought items
boughtbefore_color Numerical Average ratio of matching colors with the last three previously bought items
boughtbefore_brand Numerical Average ratio of matching brands with the last three previously bought items
boughtbefore_cat_{1,2,3,4} Numerical Average ratio of matching categories with the last three previously bought items (breadcrumb navigation level 1, 2, 3, or 4)
discount Nominal Knowledge about a discount: yes/no/unknown
discount_level Numerical Level of the discount (-1:unknown, 0:none, 1:low, 2:medium, 3:high)
viewed_before Bool Has the recommended item been viewed before
viewed_before_count Numerical Counter of previous item views
viewed_before_days_min Numerical Shortest distance to a previous item view event in days
viewed_before_days_max Numerical Longest distance to a previous item view event in days
viewed_before_sessions_min Numerical Shortest distance to a previous item view event in sessions
viewed_before_sessions_max Numerical Longest distance to a previous item view event in sessions
rec_before Bool Has the recommended item been recommended before
rec _before_count Numerical Counter of previous item recommendations
rec_before_days_min Numerical Shortest distance to a previous item recommendation event in days
rec _before_days_max Numerical Longest distance to a previous item recommendation event in days
rec _before_sessions_min Numerical Shortest distance to a previous item recommendation event in sessions
rec _before_sessions_max Numerical Longest distance to a previous item recommendation event in sessions
avg_colors_session{1,2,3} Numerical Average number of colors in the last 1, 2, or 3 sessions
avg_brands_session{1,2,3} Numerical Average number of brands in the last 1, 2, or 3 sessions
avg_price_session{1,2,3} Numerical Average number of price levels in the last 1, 2, or 3 sessions
avg_cat{1,2,3,4}_session{1,2,3} Numerical Average number of categories (breadcrumb navigation level 1, 2, 3, or 4) in the last 1, 2, or 3 sessions
user_avg_colors_session Numerical Session-wise average of different colors over all past user sessions
user_avg_brands_session Numerical Session-wise average of different brands over all past user sessions
user_avg_price_session Numerical Session-wise average of different price levels over all past user sessions
user_avg_cat{1,2,3,4}_session Numerical Session-wise average of different categories (breadcrumb navigation level 1, 2, 3, or 4) over all past user sessions
user_price Numerical Average price level of the user regarding all past actions
user_pricelevel_{view,buy} Numerical Average price level of the user regarding all past view or buy actions
user_discount Numerical Average discount level of the user regarding all past actions
user_pricereduction_{view,buy} Numerical Average discount level of the user regarding all past view or buy actions
user_viewedbefore_click Numerical Ratio of the user clicking on already known recommendations
user_viewedbefore_click_count Numerical Average number of previous item views before clicking on a recommendation
user_viewedbefore_success Numerical Ratio of the user clicking already known recommended items and buying them later on in the same session
user_viewedbefore_success_count Numerical Average number of previous item views before a sucessful recommendation (click and buy in the same session)

Attribute Weight
viewed_before 0.319285732
discount 0.27400931
discount_level 0.27400931
viewed_before_days_min 0.250696833
relpop_day 0.249434916
viewed_before_sessions_min 0.232314808
viewed_before_days_max 0.199104072
viewed_before_sessions_max 0.194098712
viewed_before_count 0.181407665
relpop_week 0.137586046
samebrand 0.110293337
relpop_month 0.078074378
samecat_1 0.076114783
brandratio_session2 0.057995586
samecat_3 0.056224732
neighbors_brand 0.055941421
brandratio_session1 0.047792972
samecat_2 0.045391729
relpop 0.045027843
samepriceratio 0.043357338
catratio_session2_4 0.042032805
similarity_viewed 0.040105184
similarity_session1 0.03794141
samecat_4 0.037708065
similarity_session2 0.037661605
priceratio_session1 0.037634752
catratio_session1_4 0.03713641
neighbors_category_4 0.034981366
priceratio_session2 0.034966541
neighbors_price 0.034580378
catratio_session2_3 0.034023242
neighbors_category_3 0.033066902
user_viewedbefore_click_count 0.031550399
neighbors_distance 0.031247694
catratio_session1_3 0.030338777
user_viewedbefore_success_count 0.030134895
user_pricelevel_buy 0.027972635
user_viewedbefore_click 0.027036008
catratio_session2_2 0.023494121
boughtbefore_brand 0.021220218
neighbors_color 0.021144145
catpop_day_4 0.020926975
catratio_session1_2 0.020686082
catratio_session2_1 0.019308092
colorratio_session2 0.01910167
catratio_session1_1 0.018785411
colorratio_session1 0.018654223
samecolor 0.01661517
user_viewedbefore_success 0.015140794
user_pricereduction 0.014332215
user_avg_brands_session 0.014134306
brandpop 0.011845966
user_pricelevel_recclick 0.011525109
user_avg_cat3_session 0.010697043
avg_pricelevels_session3 0.010582629
user_avg_colors_session 0.010155244
catpop_3 0.009875395
catpop_4 0.009474675
avg_colors_session2 0.009007016
user_pricelevel 0.008686511
catpop_2 0.008435603
catpop_1 0.008038141
avg_pricelevels_session2 0.007542855
boughtbefore_sim 0.007428123
avg_cat3_session2 0.006692441
avg_brands_session2 0.006517872
avg_cat4_session2 0.006093488
avg_brands_session3 0.005709068
colorpop 0.005185595
avg_pricelevels_session1 0.004790612
boughtbefore_cat_3 0.004721775
avg_colors_session1 0.004424866
avg_colors_session3 0.004242227
user_pricereduction_buy 0.003728988
boughtbefore_color 0.003518194
avg_brands_session1 0.002964918
avg_cat3_session1 0.002951031
avg_cat4_session1 0.002667705
avg_cat3_session3 0.001501337
user_pricereduction_recclick 0
user_pricereduction_view 0

Attribute Weight
relpop_day 1
viewed_before_sessions_min 0.624384082
viewed_before_days_min 0.619302485
relpop_week 0.610374498
viewed_before_count 0.603262067
viewed_before_sessions_max 0.597902516
viewed_before_days_max 0.595472995
viewed_before 0.590180886
discount 0.569211516
discount_level 0.569211516
relpop_month 0.371108344
brandratio_session1 0.357215418
brandratio_session2 0.351900273
priceratio_session1 0.24271177
similarity_viewed 0.233630254
samebrand 0.222211262
priceratio_session2 0.220787891
catratio_session1_4 0.210721599
catratio_session2_4 0.209751786
similarity_session1 0.204415233
similarity_session2 0.200215453
user_viewedbefore_click 0.186324741
user_viewedbefore_click_count 0.185028799
neighbors_distance 0.180883802
relpop 0.175956074
catratio_session1_3 0.171437325
neighbors_brand 0.167564153
catratio_session2_3 0.166214606
colorratio_session1 0.116326824
neighbors_category_4 0.112203813
user_viewedbefore_success 0.105168701
neighbors_price 0.101982604
neighbors_category_3 0.099244859
samecat_3 0.097115925
brandpop 0.087160435
colorratio_session2 0.085440147
samepriceratio 0.08298551
samecat_4 0.076465813
user_pricelevel_recclick 0.074521942
catpop_4 0.07375653
neighbors_color 0.067674015
user_avg_brands_session 0.065232348
catratio_session2_2 0.061275436
user_avg_colors_session 0.059402083
catratio_session1_2 0.056433885
catratio_session2_1 0.053517285
catratio_session1_1 0.053494669
samecat_1 0.051513234
user_viewedbefore_success_count 0.05151291
catpop_3 0.051312598
user_pricereduction 0.051253033
catpop_day_4 0.049767851
user_avg_cat3_session 0.049179281
samecat_2 0.043813695
boughtbefore_brand 0.043512266
catpop_2 0.037611349
boughtbefore_sim 0.033815066
samecolor 0.033796248
catpop_1 0.03039608
user_pricelevel 0.027983244
avg_pricelevels_session1 0.015262635
avg_brands_session3 0.014321607
avg_brands_session1 0.014018796
avg_cat4_session2 0.011315335
avg_colors_session1 0.01115685
boughtbefore_cat_3 0.010928953
user_pricereduction_buy 0.010682072
avg_pricelevels_session2 0.009974383
avg_colors_session2 0.009919709
avg_cat4_session1 0.009224087
avg_brands_session2 0.008910792
avg_cat3_session2 0.008084398
avg_pricelevels_session3 0.006878715
avg_colors_session3 0.006203864
boughtbefore_color 0.006059486
colorpop 0.005077786
user_pricelevel_buy 0.004557604
avg_cat3_session3 0.004381937
avg_cat3_session1 0.002760401
user_pricereduction_recclick 0
user_pricereduction_view 0

Attribute Weight
relpop_day 0.308882023
viewed_before 0.197731889
viewed_before_count 0.197731889
viewed_before_days_max 0.197731889
viewed_before_days_min 0.197731889
viewed_before_sessions_max 0.197731889
viewed_before_sessions_min 0.197731889
discount 0.190706394
discount_level 0.190659517
relpop_week 0.183649051
relpop_month 0.103124202
brandratio_session2 0.090542435
brandratio_session1 0.078734393
samebrand 0.07444879
similarity_viewed 0.057464688
priceratio_session2 0.053857527
catratio_session2_4 0.052743981
priceratio_session1 0.051899842
similarity_session1 0.050701552
neighbors_brand 0.050602911
similarity_session2 0.048551559
catratio_session1_4 0.047674563
neighbors_distance 0.043714997
relpop 0.040505645
catratio_session2_3 0.040451431
catratio_session1_3 0.037661972
user_viewedbefore_click_count 0.035530748
neighbors_category_4 0.033534052
samecat_3 0.032537339
user_viewedbefore_click 0.031553936
neighbors_category_3 0.028628693
samepriceratio 0.02780314
neighbors_price 0.026866029
samecat_4 0.025618806
colorratio_session1 0.021661597
colorratio_session2 0.019671455
neighbors_color 0.019409648
samecat_1 0.017258792
samecat_2 0.014679168
boughtbefore_brand 0.013814633
catratio_session2_2 0.012698164
catratio_session1_2 0.012084315
samecolor 0.011322963
catpop_day_4 0.010684583
catratio_session1_1 0.010597878
brandpop 0.010003377
catratio_session2_1 0.009976864
user_pricelevel_recclick 0.008124799
user_avg_brands_session 0.006529553
user_avg_colors_session 0.005370809
catpop_3 0.004711007
catpop_2 0.00468701
user_avg_cat3_session 0.004636517
catpop_1 0.00455432
boughtbefore_sim 0.004475736
catpop_4 0.004417023
avg_pricelevels_session1 0.003329095
user_pricereduction 0.003243647
boughtbefore_cat_3 0.00288359
avg_brands_session3 0.002365979
avg_brands_session1 0.002359173
avg_brands_session2 0.002326068
avg_cat4_session2 0.001781779
avg_pricelevels_session2 0.001645411
avg_colors_session2 0.001644066
avg_colors_session1 0.001643642
avg_pricelevels_session3 0.001518407
avg_cat3_session2 0.001404777
user_pricelevel 0.001393034
boughtbefore_color 0.001287128
colorpop 0.001100422
avg_cat3_session3 0.001087104
avg_cat4_session1 0.001055445
avg_cat3_session1 0.000940
user_pricelevel_buy 0.000804
avg_colors_session3 0.000792
user_pricereduction_buy 0.000689

Attribute Weight
relpop_day 0.617149564
relpop_week 0.366438849
viewed_before_sessions_min 0.340983053
viewed_before_days_min 0.337346148
discount 0.326217945
discount_level 0.326217945
viewed_before_count 0.325534402
viewed_before_sessions_max 0.319252154
viewed_before_days_max 0.317701431
viewed_before 0.312314895
relpop_month 0.218435688
brandratio_session1 0.1854455
brandratio_session2 0.182770288
priceratio_session1 0.12421509
similarity_viewed 0.11942819
priceratio_session2 0.112656574
samebrand 0.110251809
catratio_session2_4 0.108257929
catratio_session1_4 0.107926899
similarity_session1 0.103813464
similarity_session2 0.10317256
relpop 0.099561245
user_viewedbefore_click 0.099481494
user_viewedbefore_click_count 0.099316416
neighbors_distance 0.092369018
catratio_session1_3 0.088514782
catratio_session2_3 0.086324623
neighbors_brand 0.083224805
colorratio_session1 0.05822681
neighbors_category_4 0.055240741
user_viewedbefore_success 0.052715154
neighbors_price 0.050557527
neighbors_category_3 0.049113304
samecat_3 0.048016196
brandpop 0.044754316
colorratio_session2 0.042866093
samepriceratio 0.040642071
user_pricelevel_recclick 0.039465368
catpop_4 0.037916584
samecat_4 0.037302718
user_avg_brands_session 0.033699684
neighbors_color 0.033069561
catratio_session2_2 0.031496723
user_avg_colors_session 0.031042708
user_viewedbefore_success_count 0.029645357
catpop_day_4 0.029451757
catratio_session1_2 0.029362241
user_pricereduction 0.027941082
catratio_session1_1 0.027815031
catratio_session2_1 0.027598847
samecat_1 0.027213828
user_avg_cat3_session 0.026161694
catpop_3 0.025972309
samecat_2 0.022056414
boughtbefore_brand 0.021761257
catpop_2 0.0186429
boughtbefore_sim 0.016967119
samecolor 0.016402374
catpop_1 0.014987997
user_pricelevel 0.014687362
avg_pricelevels_session1 0.007469173
avg_brands_session3 0.007405807
avg_brands_session1 0.006825925
avg_cat4_session2 0.005870648
avg_colors_session1 0.005503006
user_pricereduction_buy 0.005381816
boughtbefore_cat_3 0.005302862
avg_colors_session2 0.005123122
avg_pricelevels_session2 0.004887073
avg_cat4_session1 0.004605516
avg_brands_session2 0.004461171
avg_cat3_session2 0.004062337
avg_pricelevels_session3 0.003451103
avg_colors_session3 0.003161996
boughtbefore_color 0.002940869
colorpop 0.002578514
user_pricelevel_buy 0.002532592
avg_cat3_session3 0.002121615
avg_cat3_session1 0.001337823
user_pricereduction_recclick 0
user_pricereduction_view 0