SyriaTel are a telecommunications company based in the USA who are looking to better understand the reasons behind their customer churn so they formulate strategies to better counter it. Attracting new customers can be very expensive as seems that being proactive by retaining your customers is the key to effective churn management in a telecom company
1.To build machine learning models that will predict how likely a customer will churn.
2.To find the best machine learning model for the correct classification of churned/retained customers.
3.To determine which features affect the customer churn rate thereby giving necessary recommendations.
Data containing 3333 rows and 21 columns has been provided which is more than sufficient and robust enough to be used for our analysis.That might however change if the need for more data arises.
state: 2-letter code of the US state of customer residence
account length: Number of months the customer has been with the current telcom provider
area code: 3 digit area code.
international plan: (yes/no). The customer has international plan.
voice mail plan: (yes/no). The customer has voice mail plan.
number vmail messages: Number of Voice Mail Messages
total day minutes: Total minutes of day calls.
total day calls: Total number of day calls.
total day charge: Total charge of day calls
total eve minutes: Total minutes of evening calls.
total eve calls: Total number of evening calls.
total eve charge: Total charge of evening calls.
total night minutes: Total minutes of night calls.
total night calls: Total number of night calls.
total night charge: Total charge of night calls.
total intl minutes: Total minutes of international calls.
total intl calls: Total number of international calls.
total intl charge: Total charge of international calls
customer service calls: number of calls made to customer service
churn: Customer Churn, True means churned customer, False means retained customer
With this information, the telephone company can take steps to reduce churn and retain customers. For example, they could investigate whether their international plans are meeting the needs of their customers or if there are alternative plans that may be more suitable. The company could also evaluate the importance of lowering day charges for customers, and consider the impact on churn rates if such plans are not offered. Finally, they may want to improve on their customer service to decrease the likelihood of churn.
Through various research, it has emerged that churn in the telecom industry is most often due to bad customer service. Customers canceled their contracts for the following reasons:
companies wasted their time meaning it took forever to have their issues resolved. they had to call more than once. untrained or incompetent agents. inferior self-service options. It seems that being proactive is the key to effective churn management in telecom.
See the full analysis in the Jupyter Notebook or review this presentation.