Options
Who will reply to my promotion
simone_gilardon
Member Posts: 1 
Contributor I
Hello everybody,
I'm new in the forum and in datamining also :-[
I'm working with our CRM department to identify a list of customer that will replay to a promotion
starting from DWH and CRM data (no external data at the moment) I built a csv file where I've:
CUST_ID CUST_GENDER_ID AGE_RNG TRANS_RNG FRQ_RNG OTH_PROMO REC_RNG TARGET_SS15
1009860 F 1 1 1 1 1 0
1009931 F 3 1 1 1 1 1
1015277 F 2 2 3 0 2 0
108887 F 2 1 1 1 1 0
108898 F 2 1 1 1 2 1
108930 F 1 2 2 0 1 0
108931 F 1 1 1 0 3 0
108958 F 3 1 1 1 1 0
121794 F 3 1 1 1 2 0
121841 F 2 2 1 1 2 0
121853 F 2 1 1 1 1 0
121902 F 3 1 1 0 1 0
121906 F 1 1 1 0 1 0
121945 M 2 1 1 1 1 1
121955 F 1 1 1 1 1 0
122044 M 0 1 1 1 1 0
122109 F 2 1 1 1 1 0
CUST_ID represent the customer ID
CUST_GENDER_ID Male / Female / - (at the moment I decided to keep also the non classified gender)
AGE_RNG The age has been divided in 4 classes
TRANS_RNG The number of transactions has been divided in 4 classes
FRQ_RNG The frequency the Customer visit a store has been divided in 4 classes
OTH_PROMO 0 means never replied to any promotion - 1 the customer replied in past to at least one promo
REC_RNG The recency has been divided in 4 classes
TARGET_SS15 This is the target I guess to identify... who'll reply on the next promotion based on the other data
I'm wondering which ways could I use to get the result, I was thinking to use Naive Bayesian but really I don't know if it could work and also how to use it.
any help or suggestion will be appreciated
thanks in advance
Simone
I'm new in the forum and in datamining also :-[
I'm working with our CRM department to identify a list of customer that will replay to a promotion
starting from DWH and CRM data (no external data at the moment) I built a csv file where I've:
CUST_ID CUST_GENDER_ID AGE_RNG TRANS_RNG FRQ_RNG OTH_PROMO REC_RNG TARGET_SS15
1009860 F 1 1 1 1 1 0
1009931 F 3 1 1 1 1 1
1015277 F 2 2 3 0 2 0
108887 F 2 1 1 1 1 0
108898 F 2 1 1 1 2 1
108930 F 1 2 2 0 1 0
108931 F 1 1 1 0 3 0
108958 F 3 1 1 1 1 0
121794 F 3 1 1 1 2 0
121841 F 2 2 1 1 2 0
121853 F 2 1 1 1 1 0
121902 F 3 1 1 0 1 0
121906 F 1 1 1 0 1 0
121945 M 2 1 1 1 1 1
121955 F 1 1 1 1 1 0
122044 M 0 1 1 1 1 0
122109 F 2 1 1 1 1 0
CUST_ID represent the customer ID
CUST_GENDER_ID Male / Female / - (at the moment I decided to keep also the non classified gender)
AGE_RNG The age has been divided in 4 classes
TRANS_RNG The number of transactions has been divided in 4 classes
FRQ_RNG The frequency the Customer visit a store has been divided in 4 classes
OTH_PROMO 0 means never replied to any promotion - 1 the customer replied in past to at least one promo
REC_RNG The recency has been divided in 4 classes
TARGET_SS15 This is the target I guess to identify... who'll reply on the next promotion based on the other data
I'm wondering which ways could I use to get the result, I was thinking to use Naive Bayesian but really I don't know if it could work and also how to use it.
any help or suggestion will be appreciated
thanks in advance
Simone
0
Answers
Am i right that you have historic data and that the attribute TARGET_SS15 codes whether the person replays or not? If so, you have the perfect data for supervised learning. Naive Bayes is definitly an option for this and a good baseline model. However for better results i would recommend more sophisticated methods like SVM or Random Forest.
If you have no experience with data mining at all i would recommend to have a look at this free ebok: https://rapidminer.com/wp-content/uploads/2013/10/DataMiningForTheMasses.pdf
For further more detailed studies i would recommend this book: http://www.amazon.com/Predictive-Analytics-Data-Mining-RapidMiner/dp/0128014601/ref=sr_1_1?ie=UTF8&;qid=1435305346&sr=8-1
Cheers,
Martin
Dortmund, Germany