2025年智能手环使用现状及产品发展分析报告-中南财经政法法大学

微信扫码，打开到小程序

中南的后裔 指导教师：张海波 成 员：万文娇 孔慧琳 魏羽璠 黄 菡 智能手环使用现状及产品发展分析报告 ——基于电商平台消费者购买评价的文本挖掘与武汉市的抽样调查 I 摘要 随着人工智能技术不断取得新的突破，一系列智能生活硬件逐步走进普通大 众的日常生活中。智能手环凭借入门级的售价和产品方案的标准化，市场渗透率 在智能可穿戴设备领域中一马当先，远远超过智能手表、智能眼镜等其他产品。 智能手环作为新兴产品，其市场潜力还未完全释放，正处于承前启后的发展阶段。 现阶段消费者对智能手环的评价如何？智能手环厂商今后应采取什么样的发展 策略？针对第一个问题， 我们通过抓取电商平台评论文本，使用大数据挖掘的 方法，了解消费者对产品的好评度以及关注点。针对第二个问题，通过实地调查 的方式，结合现有用户使用情况和非用户潜在价值，对智能手环未来发展策略提 出有针对性的可取建议。 考虑到在当今的移动互联网时代，通过电商平台进行网络购物已成为人们生 活中一种普遍的消费方式。利用文本挖掘的方法通过爬取天猫平台和京东平台上， 小米手环的全部可抓取评论数据，共计 468100 条。经过数据清洗后，使用中文 分词、构建语义网络与 LDA 主题模型等算法和模型，实现对文本评论数据的情 感倾向性判断以及所隐藏信息的挖掘与分析，以有效、准确的从互联网电商评论 数据中分析出消费者对小米手环的好评度、关注点以及抱怨点，对小米手环的功 能、优缺点进行初步的评价，为线下问卷调查提供方向和思路。 在问卷调查实施中，我们进行了试调查和正式调查两轮调查。在正式调查中， 我们以武汉市社区居民为调查对象，采用分层抽样和三阶段抽样相结合的概率抽 样调查方式，将武汉市的分为中心城区和远城区两层，其中三阶段抽样是从武汉 市所有行政区中利用 PPS 抽样法抽取行政区，再从行政区中抽取社区，最后利 用系统抽样的方式抽取常住居民，保证最终单元入选样本的概率均相同。试调查 问卷各项检验均通过，问卷结构合理。正式调查数据进行信度、效度分析，并且 进行游程检验和独立性检验均通过，调查结果真实可信。 针对问卷调查的结果，我们运用描述性统计方法，分别对用户和非用户的基 本信息进行分析。借助 Logit 二元选择模型，分析得到性别、年龄、学历、兴趣 爱好、性格、运动频率是购买智能手环的显著影响因素。基于回归结果，利用 K-means 聚类的方法，定量分析潜在用户性别、年龄、学历、爱好、年收入、购 买意愿，将潜在用户分为 5 类，以定性为基础结合定量方法分析了潜在客户价值。 通过文本挖掘以及问卷调查的结果，我们得到以下主要结论：第一，智能手 环已具备一定的市场认知率，从认知到实际购买的转化率较低，功能实用性是集 II 中关注点。第二，消费者对智能手环总体的满意度差强人意，天猫和京东两个电 商网站的消费者评价数据中 19.9%的消费者对购买的智能手环给予了高度满意 的评价，10.7%的用户做出了负面消极的评价；实地进行的问卷调查，我们进一 步发现，约 61%的用户对购买的智能手环整体较为满意。第三，易用人群特征 明确，通过实地调查后的建模分析，我们发现智能手环目前的消费者主要是有健 身需求的男性，对健康检测有需求的中老年消费者市场有待开发。第四，接近 45%潜在用户有较大可能成为智能手环用户。针对调查结果我们提出了相应的营 销与产品发展策略。 关键字：智能手环；不等概率三阶段抽样；文本挖掘；LDA 主题模型；logit 二 元选择模型；聚类分析法 III Abstract With artificial intelligence technology keeping getting breakthroughs , a series of intelligent life hardware is stepping into the daily life of publics. Because of the Entry-level price of smart bracelets and the standardization of its program , the market penetration of smart wearable art equipment lead the way, far more than smart watches, smart glasses and other products. As a new product, smart bracelet is now in the exploratory stage of development, so how consumers evaluate smart bracelet ? what kind of future development strategies should smart bracelet manufacturers design? For the first question, we capture the comment text in the electronic business platform and use large data mining methods to comprehend the concerns and satisfaction of consumer. As for the second question, we combined field survey with the usage experience of existing users and characteristics of non-users to make some advices about the future development strategy of smart bracelet. Online shopping has become a general method of consumption in the era of mobile Internet. We first use text mining method ,capturing data about comment of MI- bracelet in Jingdong Mall and Tmall, and we got a total of 468,100 pieces. After data cleaning, we used the way of Chinese word segmentation and built a semantic network with LDA topic model algorithms, achieving judging the emotional tendency of those text comment data and mining hidden information, so that we can analyze the Internet electricity suppliers Reviews data effectively and accurately ,besides finding the focus points and complain about MI-bracelet, which can help us make a preliminary assessment of advantages and disadvantages of this product .Then provide direction and ideas for the further study. We conducted a pilot survey and a formal survey. In the formal investigation, we treat whole community residents in Wuhan as respondents, using stratified sampling and probability sampling mode from three-stage sampling, and we divided Wuhan into 2 different parts—— the central area and urban area .In the three stage sampling ,we first used PPS sampling method to sample administrative of Wuhan, then taken further communities from selected administrative regions. Finally, the system sampling method was used in the community blocks to sample permanent residents, ensuring that the probability of the final unit selected samples are same. items of questionnaire had passed several statistic test, which means the structure of the questionnaire was IV reasonable. Formal survey data are also tested to be reliable, and valid , besides the questionnaire passed run test and independence test as well., so the survey data are authentic. Based on the Results of the survey, we made corresponding descriptive statistical analysis which reviewed the users and those non-users basic information respectively. With analysis of LOGIT binary choice model ,we concluded that sex, age, education, interests, personality and exercise frequency are significant factors of whether to buy a smart bracelet. Then based on the regression results, we made quantitative analysis about gender, age, education, hobbies, income and purchasing intention of potential users by using clustering method which can divide potential users into five categories and identified potential users’ value. From the results of the text mining and the survey, we came to some conclusions: First, smart bracelets have had a certain market awareness, but the transformation rate from cognitive to the purchase is low, besides, practicality is a main attention. Secondly, the overall satisfaction of consumers is mediocre. 19.9% consumers of Tmall and Jingdong website have relatively high satisfaction , 10.7% of users made a negative evaluation; From the survey, we found that about 61% of the users satisfied with the products. Third, features of potential users is distinct. After analyzing, we found that main consumers of smart bracelets are men who would like to exercise, and the health monitor in the elderly market needs to be developed. Fourth, about 45% potential users have a greater chance to become a real users. And according to these results, we put forward the corresponding marketing and product development strategies. Keywords: smart bracelet; three-stage PPS random sampling ; Chinese word segmentation; sentiment analysis; LDA topic model; Logit model; cluster analysis V 目录 摘要............................................................................................................................... I 一、问题综述.......................................................................................................... - 1 - （一）研究背景............................................................................................... - 1 - （二）文献综述............................................................................................... - 3 - 二、研究思路与创新.............................................................................................. - 4 - （一）研究思路............................................................................................... - 4 - （二）研究意义............................................................................................... - 6 - （三）特色与创新........................................................................................... - 7 - 三、基于电商平台评论数据的文本挖掘.............................................................. - 9 - （一）评论数据采集....................................................................................... - 9 - （二）数据预处理......................................................................................... - 11 - （三）中文分词............................................................................................. - 13 - （四）基于情感词典的情感倾向性分析..................................................... - 12 - （五）基于 LDA 模型的主题分析................................................................- 14 - 四、调查方案........................................................................................................- 17 - （一）调查目的............................................................................................. - 17 - （二）调查对象和调查单位......................................................................... - 17 - （三）调查项目............................................................................................. - 17 - （四）调查方式与方法................................................................................. - 18 - （五）抽样调查基本步骤............................................................................. - 20 - 五、调查实施.......................................................................