商品星力评级模型数据

在电商人货场三个数据要素中, 产品起着至关重要的作用，行业内有句流行话是这么说的, 选品定生死。因此， 如何快捷的把产品分类，区分出引流款、利润款、爆款、形象款等， 需要一个包含成交能力、流量获取、流量转化、流量价值、客户拉新、质量服务的综合性多维度的商品星力评级模型。采用商品星力评级模型对商品进行价值评级，参考经典的客户RFM分析方法，把商品计算出一个综合性的星力等级分数，分值越小，等级越高，对细分过后的不同商品采取相应营销策略，进行精准有效的运营。通过对商品进行分级管理，满足不同等级的商品个性化需求，并为同行业企业管理不同等级的商品，实现精准个性化服务提供数据支持，具有显著的经济效益和社会价值。处理：从天猫-生意参谋-商品中商品效果基础数据，对数据脱敏、降噪、清洗、聚集。2、加工：对商品做一个综合排名，得出一个商品星力总评分。 a.成交能力，对支付金额进行排名，支付金额最大商品排在最上面。按照从1-5评分，根据支付金额_排名_百分比排名分位，前20%的商品获得1分，接下来的20%用户获得2分,再下来20%的商品为3分，再下来20% 的商品为4分，最后20% 的商品为5分。 b.流量获取，根据商品访客数，前20%的商品的分数为1，以此类推。 c.流量转化，根据支付转化率，前20%的商品分数为1，以此类推。d.流量价值，通过访客平均价值最高的20%商品则分数为1。e.客户拉新，商品拉新=0.5*支付新买家数占比+0.5*支付新买家金额占比，商品拉新前20%的分数为1，以此类推。f.质量服务,按照退款率，进行升序排名，规则前述类似。商品星力评级=0.3*成交能力+0.25*流量获取+0.1*流量转化+0.1*流量价值+0.1*客户拉新+0.15*质量服务 ，评分小于2分的为A级，大于等于2小于等于3为B级，大于等于3小于4的为C 级，大于4的为D 级， A级为最高等级。

Among the three core data dimensions of e-commerce: people, product and scenario, products play a pivotal role. There is a well-known saying in the industry: "Product selection makes or breaks the business". Therefore, to quickly classify products and distinguish them into drainage products, profit products, best-selling products, image products and other categories, a comprehensive multi-dimensional Product Star Rating model covering transaction capability, traffic acquisition, traffic conversion, traffic value, customer acquisition and quality service is required.

Using the Product Star Rating model to evaluate product value, drawing on the classic customer RFM analysis framework, a comprehensive star rating score is calculated for each product. The lower the score, the higher the corresponding product rating. Corresponding marketing strategies are adopted for classified products to achieve precise and efficient operation.

By implementing graded product management, this model meets the personalized needs of products at different levels, and provides data support for enterprises in the same industry to manage products of varying tiers and deliver precise personalized services, yielding substantial economic and social benefits.

### Data Processing Workflow
1. Preprocessing: Extract basic product performance data from the Product module of Tmall Business Consultant, followed by data desensitization, denoising, cleaning and aggregation.
2. Scoring & Ranking: Conduct comprehensive ranking for products to derive the overall Product Star Rating score.
a. Transaction Capability: Rank products by their payment amount, with the product with the highest payment amount ranked first. Assign scores from 1 to 5 based on the percentile rank of payment amount: the top 20% of products receive 1 point, the next 20% receive 2 points, the subsequent 20% receive 3 points, the following 20% receive 4 points, and the last 20% receive 5 points.
b. Traffic Acquisition: Rank products by their visitor count, with the top 20% receiving 1 point, and so on for the remaining percentile groups.
c. Traffic Conversion: Rank products by their payment conversion rate, with the top 20% receiving 1 point, and so on.
d. Traffic Value: Rank products by their average visitor value, with the top 20% (highest average visitor value) receiving 1 point, and so on.
e. Customer Acquisition: Calculate the customer acquisition score as 0.5 * proportion of new paying buyers + 0.5 * proportion of transaction amount from new paying buyers. Rank products by this customer acquisition score, with the top 20% receiving 1 point, and so on.
f. Quality Service: Rank products by their refund rate in ascending order, applying the same percentile-based scoring rule as above.

The overall Product Star Rating is calculated as:
Product Star Rating = 0.3 * Transaction Capability Score + 0.25 * Traffic Acquisition Score + 0.1 * Traffic Conversion Score + 0.1 * Traffic Value Score + 0.1 * Customer Acquisition Score + 0.15 * Quality Service Score.

The rating levels are defined as follows:
- Level A (highest rating): scores below 2
- Level B: scores ≥2 and ≤3
- Level C: scores ≥3 and <4
- Level D: scores >4

搜集汇总

背景与挑战

背景概述

该数据集是一个电商商品评级模型数据，包含673条记录，每月更新，用于通过多维度评分模型对商品进行综合评级，支持精准营销和商品管理。