0
$\begingroup$

Is it readily possible to do predictive churn analysis (i.e., associating a churn risk with every individual/customer) using statistical tools (e.g. in Excel) not involving the use of machine learning or AI or deep learning? Is such a thing commonplace at all?

My guess was that people would be doing some sort of predictive modeling (for instance using some sort of cohort analysis and/or multiple regression) in this area before ML came on to the scene, but my online search, including on these forums, yields me nothing of the sort.

Does there exist a relatively simple churn risk analysis without ML that is as simple as practical while still having actual business utility? If such a thing is described anywhere, could you please point me in the right direction? The same thing could be applied to similar scenarios such as employee attrition. Any guidance will be appreciated. Thanks a bunch.

[CLARIFICATION IN RESPONSE TO BOT COMMENT]

Let's say we have 200 to 2000 records that tell us about customers who have subscribed to our services. It tells us when they subscribed, when they left (if they did), what particular services they are subscribed to, what they are paying, etc. It also tells us how old they are, roughly how much they earn, and perhaps some other demographic info that can help us classify them into groups of cohorts of some sorts (e.g. young single urban males, married female business owners, etc.)

With this size of data (certainly not enough to "train" a machine learning model - nor is that my goal), do there exist analytical models that help approximate the churn risk for particular individuals - at any level of granularity? Are examples/illustrations of this available online? I am willing to use statistical tools such as multiple regression if really required but I want to do it purely in Excel, without deploying any machine learning algorithms. Hope this clarifies.

$\endgroup$
5
  • $\begingroup$ Please clarify your specific problem or provide additional details to highlight exactly what you need. As it's currently written, it's hard to tell exactly what you're asking. $\endgroup$
    – Community Bot
    Aug 4 at 12:56
  • $\begingroup$ Can you clarify what you mean by "without ML"? By the most commonly used definitions (e.g. something along the lines of "method for making a computer do some task after somehow training on some data") regression models are a form of ML. By the way, Googling something like +"churn risk" +"logistic regression" seems to produce a lot of hits, did you do something like that? $\endgroup$
    – Björn
    Aug 4 at 13:23
  • $\begingroup$ @Björn Appreciate your query. To clarify, I am making a distinction between a purely computational approach that can be purely defined and solved as mathematical problem using Excel, and a machine learning approach which involves use of specialized software, e.g. those using neural nets, for "training" the software in making connections between inputs and outputs. I'd neither have the software for that, nor the "big data" for training it. Only actual business data (may be 10's to 1000's of records for mathematical analysis, but not the millions required for typical "ML"). Hope this clarifies? $\endgroup$ Aug 4 at 20:05
  • $\begingroup$ @Björn To answer your second question, I googled it and only found results that use machine learning algorithms - nothing purely in Excel (or even Power BI/Power Query etc.) $\endgroup$ Aug 4 at 20:14
  • $\begingroup$ @Community Added details of specific business problem $\endgroup$ Aug 4 at 20:15

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.