Cognub has designed a Recommendation Engine, which provides recommendation for the best fit car to a consumer considering the trends in the demographic and market segments. Recommendations are personalized to meet the expectation based on purchase preferences.
With the online car trading being a new pastureland, the technique has gained importance to target fruitful recommendation towards users. The system proposed recommends best fit items considering the attributes not limited to those such as item complexity, purchase behavior, impact of the item in the market, effect on the relevant social segments and similarity of the listed item to those in the inventory.
The technique can be realized in three broad steps:
- Scoring the individual cars based on the attributes like MSRP, Mileage, Engione Displacement, Safety Features, Interior features and similar features that influence the vehicle preferences.
- Popular Trim identification: This could be either done with the help of a expert survey report on the basis of the sale statistics, or could be derived inherently with the help of a collaborative filter. Either way the collected information will add onto the collected score.
- Scoring the trim attributes is a challenge due to its high variability in the description method which can be overcome using Tanimoto scoring on an NLP perspective.
The appropriate mixture of the above three scores are used to rate the car on the scale of 1-10, that would be stored as a template. The recommendation is then made by calculating a distance metric of the listed car with the others on the proprietary score mentioned above.