I have a general question about how to tackle an analysis task. The goal is to understand which store needs which cleaning services to increase customer satisfaction with the cleanliness of the specific store.

I have different data about several thousand stores:

  1. the store properties (I think the term is master data) like store
    area, store location (urban/countryside), building complexity(high/middle/low), tourism hotspot (yes/no), etc. These data are mostly static and cant be changed easily.
  2. data about how much and which cleaning service was performed at this store like windows cleaning (30m² - 10 times/month), floor cleaning, etc.
  3. and the target parameter which is the grade the customer gave for the cleanliness of the store (1-100)

My specific question is if it makes sense to take the master data and classify my stores into different clusters and just then use the regression models on every cluster separately. I would then only use the data about the cleaning services as features and not the master data anymore since I used them for the classification already. I ask because, I assume that some stores are so vastly different from other stores that they also need totally different cleaning services.

I'm not that experienced with analyzing data so I hope my question/approach makes sense.

Edit: To clarify the intent more. Every store has a budget for cleaning. The cleaning services are ordered from a third party but we tell them which cleaning services we need. So if we tell them to sweep they won't touch other dirty areas at all. I want to know which cleaning services are actually the most noticeable by the customer. For example if the expensive ceiling cleaning is even worth it or maybe we should invest more in high-pressure cleaning of the floor etc.

We can always just pump more money into the cleaning services but we made the negative experience with the impact of just increasing the budget without having a clear understanding of what the customers actually want to have cleaned. So now I want to understand the impact of the cleaning services given the store environment better and do informed investments.

  • $\begingroup$ Why are you running a regression model at all? Isn't it obvious from the customer satisfaction scores which stores need more cleaning? If you explain how you expect a model to help make your decision then it would be much easier to advise how you should go about constructing it. $\endgroup$ Mar 5, 2020 at 10:38
  • $\begingroup$ We want to clean more in the "dirty" stores but make smarter investments. Ive added more context in the edit. $\endgroup$ Mar 5, 2020 at 18:10


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.