top of page
CONCLUSION
CONCLUSION
insights
Insights/Inferences from Data Visualizations Plots
Data Visualization Box Plot for Customer Service Calls
We see that the number of customer service calls made to customers who churn is relatively high.
This indicates that customers who have churned have tried contacting customer service
but might have not received a satisfactory resolution to their issue.
Data Visualization Plot for International Plan
This variable shows a much more meaningful relationship. Roughly 10% of the customers which have international plan , and overall out of those approx 42% of customers churn.
On the other hand, of the 90% of customers that does not have international plan , out of those only 12% churn. We analyze that those customer not having international plan are most likely not churn.
Data Visualization Plot for Voice Mail Plan
This variable also shows a much more meaningful relationship. Roughly 28% of the customers which have VMail plan , and overall out of those approx 9% of customers churn.
On the other hand, of the 72% of customers that does not have VMail plan , out of those only 17% churn. We analyze that those customer not having VMail plan are most likely not churn.
.
Data Visualization Density Plots for Day Charge,Eve Charge,Night Charge,Intl Charge weighted by churn variable
#for Night Charge the density of customers which are more likely to churn and those which are less likely to churn is quite symmetrical in nature.
#for Day Charge the density of customers which will churn and which will not churn is not symmetrical in nature and also the customers which are more likely to churn are
having higher Day Charge as compare to customers which are less likely to churn,
howsoever they have lower density levels.
#for Eve Charge the density of customers which are more likely to churn and those
which are less likely to churn is nearly similar in nature,as well as there density levels
#for Intl Charge the density of customers which will churn and which will
not churn is not symmetrical in nature and also the customers which
are more likely to churn are having higher Intl Charge as compare to
customers which are less likely to churn,howsoever they have lower density levels.
.
Data Visualization Box Plot for Voice Mail Message
The number of voicemail messages seems to be lower for customers who churn compared to those who don't.
Data Visualization Weighted Scatter Plot between Customer Service Calls and Day Calls,Eve Calls,Night Calls,Intl Calls respectively
We analyze that relatively when the number of customer service calls reaches 4 and
above against day,eve,night,intl calls ,there are more chances of that the particular customer
is more likely to be a churn customer and below 4 relatively with the number of customers
service calls it is a customer which is less likely to churn.
Data Visualization Weighted Scatter Plot between Customer Service Calls and Day Charge,Eve Charge,Night Charge,Intl Charge respectively
We analyze that relatively when the number of customer service calls reaches 4 against day,eve,night,intl charge and more,there are more chances of that the particular customer
is more likely to churn and below 4 relatively with the number of customers service calls
it is a customer which is less likely to churn and also we see day charge is higher within
when no of service calls is less than 4 and later in comes down relatively when
comparing it with night,eve,intl charges.
Data Visualization Box Plot between Churn and Day min,Eve min,Night min,Intl min repectively
We analyze that the total number of day minutes for customers likely to churn are higher
as compare to customers which are not churn and also the total number of minutes remain
relatively same for eve,night,intl customers.
CONCLUSION
Predicting whether a customer will stop using your product or service is an important component of customer behavior analytics.The big challenge these days is to increase customer loyalty before subscribers decide to leave and to aim
efforts at customers who are at risk of churning.For large companies it becomes very difficult to analyse every cases, one
by one, to understand the cause out of that which customer churn/not churn.But what if we can predict which customer will churn.Statistical Analysis has given us the power to predict the probability of an customer which can churn by virtue of
fitting a well selected and accurate model.This can be an enormously useful information for the company perspective
which will allow them to figure out which customer is more likely to churn and thus in advance help them in reducing the risk involved while associating with that customer.If the company predicts the churn rate of the customers with high accuracy,it gives the company a estimate of how its revenues would look like and in turn give it freedom to plan finances ahead.
Insights From the Model and Business Recommendations to Company based on the discriminant coefficients and the correl_ratio provided by the model, an increase in the below variables increases the probability of customer churn:
Number of customer service calls,Total day charge,Total evening charge,Total international charge,Total night charge
Additionally, an increase in number of voice mail messages decreases the probability of customer churn.
At the end,Day Calls and Day Charge,One could argue that both are relevant,as an consumer might churn as
making many calls turns out to be problematic(bad signal, quality for example). While you could also argue the
price will determine the churn.
These insights from the discriminant model can help the business/company formulate strategies to reduce customer churn.
Here's what I would recommend to the business/company based on what we've learned: customer issues should be
resolved within the three or fourth call, as repeated calls to customer service causes customer churn there should be an organized escalation procedure for issues not resolved within four calls the provider should offer more attractive plans that reduce the cost of day, evening, and international calls based on usage.
Hence There are a number of other different insights that we could gain from the data, but this would be a good
initial list to investigate further if the company had even more detailed data sets.
bottom of page