創新性與消費行為資料分析 - 以台灣流行服飾產業為例

Abstract

在日常生活當中，每天都有新的產品推出，而新產品對於企業的營收貢獻隨著科技發展也越來越重要，在流行服飾產業中尤是。根據創新擴散曲線，新產品要成功被銷售，需要由創新消費者作為市場擴張的開端。過往定義消費者創新性時多使用自填表單或透過質化的方式，即使經過效度、信度以及實務使用上的認證，但自我表達與真實行為仍存在一段落差，因此本研究將透過數據資料找出潛藏在龐大資料庫中的創新消費者，並以RFMI的分群方法，研究各群消費者的消費行為。 研究方法是利用流行服飾產業消費者購買行為資料進行數據分析，透過限制資料在觀察時間的範圍內，確保資料能夠完整呈現結果。本研究將消費者購買產品的時間給予其創新性的計算，再透過探索式的資料分析，初步將人口統計變數、消費者購買行為、消費者瀏覽行為分析的結果視覺化並依序與創新性進行變數的關聯性分析，最後結合RFM與K-Means分群演算法，確實發現創新性對於企業營運具有相當的影響力。 研究結果顯示，高創新性消費者表現在年齡、消費通路、使用優惠行為、搜尋關鍵字、網頁瀏覽行為都與低創新性消費者有所不同。而高創新性消費者的平均消費金額也同時顯著高於低創新性的消費者，對企業的營運貢獻上非常大。因此後續若能依照創新性對消費者進行分群行銷，並對流失的顧客進行訪談，找出流失的痛點，將能夠改善企業在營運發生的問題。

In daily life, new products are launched every day, and the contribution of new products to the company's revenue is becoming more and more important with the development of technology, especially in the fashion industry. According to the diffusion of innovative theory, for new products to be successfully sold, innovative consumers need to be the beginning of the market expansion. In the past, when defining innovation, we often used self-report or through qualitative methods. Even after being validated in terms of validity, reliability, and practical use, there is still a gap between self-report and customers’ real behavior. Innovative consumers’ behavior in a huge database is the most important research topic of this research. The research method is to use the data of the customer purchasing behavior of the fashion apparel industry for data analysis and to limit the data within the observation time to ensure that the data can completely present the results. This study calculates customers’ innovativeness by using their purchasing data, and then through exploratory data analysis to analyze demographic variables, consumer purchasing behavior, and consumer browsing behavior. Analysis including visualized data, correlation analysis and finally combining RFM and K-Means grouping algorithm. As a result, the study is indeed found that innovation has a considerable influence on business operations. The results show that innovative performance is significantly different in age, search behavior, use of preferential behavior, and contribution to the company's operation. Therefore, if follow-up user interviews, focus groups, product design adjustments, and A/B tests can be very important for the business operation.