Frequency bin class

[gfagherazzi]

Hi, I need to modify the way RFM bins are created.
For frequency, for example, the 2 class is too large in my case.
Thank you

[aravindhebbali]

Hi @gfagherazzi, Is it possible to share some more information? It will help me to understand why 2 class is large.

Thank you

[gfagherazzi]

Hi @aravindhebbali , yes sure
Please have a look at my heatmap.

[aravindhebbali]

This usually happens when there is not much variation in the frequency data i.e. the transaction count for the customers. In your data set, there are customers either with very low or very high transaction count.

[gfagherazzi]

Ok I see, is it possible to read somewhere the RFM threshold for each class?
For instance, in this case, it should be very useful to know how the frequency class is split between the 5 bins.
Thank you
G.

[aravindhebbali]

Ok, I will create a new branch in which the output will include the thresholds used. Which of the below function are you using to create the RFM results?

• rfm_table_order()
• rfm_table_customer()
• rfm_table_customer_2()

I will create a variant of the function so that it will return the thresholds.

[gfagherazzi]

Wow super!
I'm using the rfm_table_order() function

[aravindhebbali]

I have created a new branch threshold. Install the package from this branch and you should be able to access the thresholds.

devtools::install_github("rsquaredacademy/rfm@threshold")

analysis_date <- lubridate::as_date('2006-12-31', tz = 'UTC')

result        <- rfm_table_order(rfm_data_orders, 
                                 customer_id, 
                                 order_date, 
                                 revenue, 
                                 analysis_date)

result$threshold
# A tibble: 5 x 6
  recency_lower recency_upper frequency_lower frequency_upper monetary_lower monetary_upper
          <dbl>         <dbl>           <dbl>           <dbl>          <dbl>          <dbl>
1            1           115                1               4            12            256.
2          115           181                4               5           256.           382 
3          181           297.               5               6           382            506.
4          297.          482                6               8           506.           666 
5          482           977                8              15           666           1489 

The intervals are created in the below style:

Left-closed, right-open: [ a , b ) = { x ∣ a ≤ x < b }

In the above example, the bins are assigned as follows:

Bins Frequency/Transaction Count
   1                       1 - 3       
   2                           4
   3                           5
   4                       6 - 7
   5                          >7

Hope this helps.

[gfagherazzi]

Super useful
Thank you so much!

G.

[gfagherazzi]

Sorry to disturb you again, but if think it should be really useful a method to override values in rfm_result tibble bins.

Something like segments modeling

recency_lower <- c(4, 2, 3, 4, 3, 2, 2, 1, 1, 1)
recency_upper <- c(5, 4, 5, 5, 4, 3, 3, 2, 1, 2)
....

Thank you
G.

[aravindhebbali]

You mean you want to be able to specify the intervals manually?

[gfagherazzi]

Yes exactly

[aravindhebbali]

#58

Will experiment with this enhancement in the threshold branch sometime this week and will keep you posted.