The Altair Community is migrating to a new platform to provide a better experience for you. In preparation for the migration, the Altair Community is on read-only mode from October 28 - November 6, 2024. Technical support via cases will continue to work as is. For any urgent requests from Students/Faculty members, please submit the form linked here
What is the predictive model to use here ?
Contributor I
Hello to everyone,
I am working right now on a project and I really need your precious help. I am working right now on a project where I have to send sales action letter to customers to propose them up-sells. To accomplish this mission, I have to define first of all the PROFILE of customers who are the most likely to accept those up-sells. I have a database that provides a single customer view; for every single customer, we have data that characterize the customer, data that concerns the different transactions he/she has had in 2011 with our company, and finally an important data that tell us the number of days that the customer took to reply to our sales action letter sent in 2011 letter after receiving it.
ASSUMPTIONS
We’re gonna assume that our company sells 5 types of products: A, B, C, D, E and those 5 different products can be sold through 3 different channels : our stores, the franchisees and other points of sales.
DATA
As said earlier, in my database, I have a single customer view, I mean I can see for every single customer the following data:
Data that characterize personnally customers:
- The Age of the customer (Variable: Age)
- The postal code of the customer (Variable :Postal Code)
- The social class (the variable social is a ordinal variable and people who rank at the top of the ranking are the wealthiest people and inversely for the people who ran at the bottom of the ranking)
Data that concerns customer transactions:
- The number of products that the customer has bought in our company (We assume that we sell 5 types of products : Product A, Product B, … Product E and there is every time one variable representing one product. At the end, we have 5 variables)
For instance – Variable A: Numerical variable that contains the number of Products A that were bought by the customer
- the number of products that the customer bought in our stores, in the franchisees and in other point of sales. (There are three variables representing the 3 different channels: Variable Stores, Variable Franchisees and Other point of sales and they count the number of products that the customer bought in each channel)
A Data that assesses the rapidity to which the customer replied positively to our sales action
- The time he/she took to reply to the up-sell (Variable: Number of days to reply to the action letter since it was sent)
QUESTION:
Based on this dataset described here above, I have to build a predictive model to predict in the future for every single customer if the customer will replied positively to our sales action or not based on the data we have on their personal characteristics and on their transactions. Which is the best predictive model to do that and why ? Decision tree ? OLS Regression ?
I was thinking about doing an OLS regression but it will give me only a general idea of the profile of the customer most likely to reply to our sales letter. It is very important to me to have a kind of ranking of customers who are most likely to reply to sales action letter.
I would be very thankful if you could help me.
Mdna
I am working right now on a project and I really need your precious help. I am working right now on a project where I have to send sales action letter to customers to propose them up-sells. To accomplish this mission, I have to define first of all the PROFILE of customers who are the most likely to accept those up-sells. I have a database that provides a single customer view; for every single customer, we have data that characterize the customer, data that concerns the different transactions he/she has had in 2011 with our company, and finally an important data that tell us the number of days that the customer took to reply to our sales action letter sent in 2011 letter after receiving it.
ASSUMPTIONS
We’re gonna assume that our company sells 5 types of products: A, B, C, D, E and those 5 different products can be sold through 3 different channels : our stores, the franchisees and other points of sales.
DATA
As said earlier, in my database, I have a single customer view, I mean I can see for every single customer the following data:
Data that characterize personnally customers:
- The Age of the customer (Variable: Age)
- The postal code of the customer (Variable :Postal Code)
- The social class (the variable social is a ordinal variable and people who rank at the top of the ranking are the wealthiest people and inversely for the people who ran at the bottom of the ranking)
Data that concerns customer transactions:
- The number of products that the customer has bought in our company (We assume that we sell 5 types of products : Product A, Product B, … Product E and there is every time one variable representing one product. At the end, we have 5 variables)
For instance – Variable A: Numerical variable that contains the number of Products A that were bought by the customer
- the number of products that the customer bought in our stores, in the franchisees and in other point of sales. (There are three variables representing the 3 different channels: Variable Stores, Variable Franchisees and Other point of sales and they count the number of products that the customer bought in each channel)
A Data that assesses the rapidity to which the customer replied positively to our sales action
- The time he/she took to reply to the up-sell (Variable: Number of days to reply to the action letter since it was sent)
QUESTION:
Based on this dataset described here above, I have to build a predictive model to predict in the future for every single customer if the customer will replied positively to our sales action or not based on the data we have on their personal characteristics and on their transactions. Which is the best predictive model to do that and why ? Decision tree ? OLS Regression ?
I was thinking about doing an OLS regression but it will give me only a general idea of the profile of the customer most likely to reply to our sales letter. It is very important to me to have a kind of ranking of customers who are most likely to reply to sales action letter.
I would be very thankful if you could help me.
Mdna
0
Answers
The other advantage is more a marketing angle - people do tend to follow crowds so telling them that X thousands of customers who bought the same item they bought also found this new exciting item Y interesting can drive + behavior.