An Effective Approach to Promote Air Traveler Repurchasing Using the Random Forest Algorithm: Predictive Model Design and Utility Evaluation

Table 1

The definitions of 28 feature variables.

Variables	Defines
Cust_type	Membership level (1–5); ranging from 1 (lowest level) to 5 (highest level) with the midpoint 3 (general level);
Status	Account status (0–2); elapsed: 0; slumber: 1; active: 2;
Is_lyreg	Is it a member account; nonmember: 0; member: 1;
Ly_status	Whether the member account is activated; nonactivated: 0; activated: 1; very active: 2
Gender	User gender; female: 0; male: 1;
Card_province	Province where the user ID is located;
Tel_province	Province where the user's mobile phone is located;
Order_buy_zh	Whether to buy economy class tickets; noneconomy class: 0; economy class: 1;
Order_language	The language preference of air tickets orders;
First_buy_diff	Difference between first purchase date and registration date;
Last_diff	Days between last two purchases;
Last_amount	The amount paid for the last purchase;
Flight_mile	Total flight distance;
Order_num_m24	Number of orders paid in the last 24 months;
Order_num_m12	Number of orders paid in the last 12 months;
Order_num_m6	Number of orders paid in the last 6 months;
Order_num_m3	Number of orders paid in the last 3 months;
Order_num_d30	Number of orders paid in the last 1 month;
Tkt_fee_all	Total payment;
Tkt_fee_m12	Total payment amount in the last 12 months;
Tkt_fee_m6	Total payment amount in the last 6 months;
Tkt_fee_d30	Total payment amount in the last 1 month;
Avg_daydiff	Average purchase frequency;
Avg_tktdiscount	Average discount;
Avg_tktfee	Average ticket price;
User_type	Type of membership (general member, silver member, gold member);
User_cycle	Member's life cycle;
Order_num_all	Total payment orders;