Predictive modeling challenge for SPSS experts.........!

[Mahendra]

Urgent! Need Help to build best Model for the problem

Hi,


I have a raw attribute data for 430 restaurants.

Training Data: The set contains information about 430 restaurants. The target variable for every row is the aggregated weekly footfall named ‘Total’. I need to build their model using this training set.

Test Data: This set has the all the attributes except the target variable (Total Guest count) information. I have to predict the target variable based on the test set


The goal is to make the appropriate model using training data set and then predict the guest count at additional restaurants given in test data (70 restaurants) set using this model and finally to make recommendations to improve footfall

 

[Eyecatcher]

Hi,

Nice exercise, but I don't understand your point. Do you want us to find the best model? Do you want any help with how to build such a model?

 

[Mahendra]

Re:Urgent! Need Help to build best Model for the problem

Hi,

Nice exercise, but I don't understand your point. Do you want us to find the best model? Do you want any help with how to build such a model?

I want help and the correct approach to build the best model for the given data.
I have used MLR technique to build my model , but RMSE which I am getting is ~185.00.

I am newbie in this field and looking for expert advice

 

[Eyecatcher]

Ok, let me start by asking a few questions:

- Why is the data aggregated? Is this the only data you have? (for instance, do you also have data about marketing actions per restaurant?)

- Did you use linear regression? And why not a multiplicative version (cobb-douglas function)? Did you consider interaction effects?

- Can you show us some output? Which variables did you find significantly contributing?

- Did you check the assumptions of OLS?

- Is this a homework assignment?

 

[Eyecatcher]

http://www.crowdanalytix.com/contes...unnchh-analytics-competition-1403201523/view/

Your deadline was yesterday huh?