Related Posts
Updated my wifi SSID

Does Coforge pays the Variable full?
Additional Posts in Data & Analytics Consultants
Has anyone else begun to resent data science?
New to Fishbowl?
Download the Fishbowl app to
unlock all discussions on Fishbowl.
unlock all discussions on Fishbowl.




Maybe something really simple like k-means clustering on a single variable to bucket them.
This assumes that you don’t want to bother looking at the distribution to figure it out.
You could also just split into quantiles. 0-20th percentile, 20th to 40th, etc.
Thanks for the ideas!
How about plotting out the cumulative number of clicks, with price advertised as the X axis? Expectation would be that it’s a curve with higher slope at lower prices and flattens out as you reach higher prices, and you could eyeball a region where the curve flattens the most rapidly as a starting range for which prices to investigate further.
Are you trying to predict a binary variable (purchased vs did not purchase) or a value (purchase amount)?
I’ll add that I question the usefulness of bucketing these variables. Maybe run a model through xgboost/random forest with price/basket size as continuous. Pull the shap values and then plot price vs shap value and see if there are any discontinuities or if it is a linear relationship.
Many ways to tackle this problem.