The experiment

We solved the optimal assortment problem if the customer selects under each of the 4 MNL-CE models. The results generated have two sets of information:

  1. How many items to recommend.
  2. No choice probability value when the recommended size of the assortment is offered.

Read in the data files to data frames

Number of items to recommend

Here we summarize the results using descriptive statistics

##     MNL_CE1l        MNL_CE1nl        MNL_CE2l      MNL_CE2nl    
##  Min.   : 14.00   Min.   : 7.00   Min.   :20.0   Min.   :20.00  
##  1st Qu.: 59.00   1st Qu.:14.00   1st Qu.:40.0   1st Qu.:40.00  
##  Median : 70.00   Median :17.00   Median :60.0   Median :40.00  
##  Mean   : 68.44   Mean   :16.43   Mean   :53.2   Mean   :38.81  
##  3rd Qu.: 79.00   3rd Qu.:20.00   3rd Qu.:60.0   3rd Qu.:40.00  
##  Max.   :100.00   Max.   :29.00   Max.   :80.0   Max.   :40.00
##     MNL_CE1l           MNL_CE1nl            MNL_CE2l        
##  Min.   :0.0001831   Min.   :0.0009119   Min.   :0.0001738  
##  1st Qu.:0.0002775   1st Qu.:0.0009149   1st Qu.:0.0002342  
##  Median :0.0003729   Median :0.0010221   Median :0.0003183  
##  Mean   :0.0004440   Mean   :0.0012390   Mean   :0.0003824  
##  3rd Qu.:0.0005008   3rd Qu.:0.0013604   3rd Qu.:0.0004325  
##  Max.   :0.0042023   Max.   :0.0063336   Max.   :0.0034588  
##    MNL_CE2nl        
##  Min.   :0.0003276  
##  1st Qu.:0.0003812  
##  Median :0.0004905  
##  Mean   :0.0005770  
##  3rd Qu.:0.0006516  
##  Max.   :0.0040929

Exploratory data analysis

Our aim is to understand the distribution of assortment sized for all users under the 4 MNL-CE models.

For the scatter plots:

Black = MNL-CE1l

Orange = MNL-CE1nl

Blue = MNL-CE2l

Red = MNL-CE2nl

Histogram for the 4 models

## Boxplots

boxplot(Datanumbertorec[c(-1)])

boxplot(Datanochoice[c(-1)])