2
$\begingroup$

I have data that has actions that perform on my tool and I would like to predict the customers who are ready to convert from free/trail to the paid category.

My data looks like the following:

dummy<-data.frame(license=sample(c("Free","Trail","Paid"),10000,replace = T,prob = c(0.6,0.35,0.05)),
           plan_type=sample(1:5,10000,replace=T),
           action1=sample(0:100,10000,replace = T),
           action2=sample(0:1000,10000,replace = T),
           action3=sample(0:10,10000,replace = T),
           num_days_in_product=sample(0:500,10000,replace = T))


head(dummy)
  license plan_type action1 action2 action3 num_days_in_product
1    Paid         1     100      71       5                 285
2    Free         5      75     438       1                   2
3    Free         1       5     555       7                 389
4    Free         3       4     105       0                 150
5    Free         1      16     348       7                 423
6    Free         5      15     866       8                 270

> table(dummy$license)

 Free  Paid Trail 
 6016   516  3468 
> prop.table(table(dummy$license))

  Free   Paid  Trail 
0.6016 0.0516 0.3468 

Let me know if any extra information needed.

$\endgroup$
2
$\begingroup$

Two items come to mind:

  1. You will gain the most benefit from this site when you propose a model and let other people comment on and modify the model. As it stands, you are merely proposing what your data looks like. Do you have a model that you've tried? What data science steps have you already taken with your project? If you post that, this site will be of much better help to you
  2. You should be aware that what you posted makes for a good initial dataset but that's all it is, your initial set. If I was in your position, I would be working towards developing a meta dataset that contains a much larger number of factors. When did the person join? What was the time in between actions? Which country is their IP address from? How many times have they logged into the tool? Etc, etc, etc. When you start to think of your data in this manner, and give yourself a much larger number of factors to start with, then you are taking steps that are much more likely to result in a reliable predictive model.
| improve this answer | |
$\endgroup$
  • 1
    $\begingroup$ It might also help to explain what action 1 2 3 and their value means conceptually so that we can help you better. Another thing I would add to the great response above is the trigger point. When did the consumer switch from a Free plan to a paid plan vice versa and the actions leading to such event would be helpful. $\endgroup$ – The Lyrist Jan 4 '19 at 17:21

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.