Implementación del método no supervisado PAM para la segmentación de clientes en una tienda virtual farmacéutica del año 2022
No Thumbnail Available
Date
2024
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad Nacional de Trujillo
Abstract
En estos tiempos, las empresas se encuentran ante el reto de gestionar volúmenes significativos de datos, optimizar su uso y convertirlos en información de gran valor que genere ideas brillantes y relevantes para la toma de decisiones y concebir estrategias beneficiosas. A lo largo de diversas disciplinas, una técnica comúnmente empleada es el análisis de agrupamientos, sin embargo, su aplicación más beneficiosa se observa en el ámbito empresarial, abordando aspectos como la segmentación o clasificación de clientes y la detección de tendencias en los patrones de compra. De esta forma, el caso de la tienda virtual farmacéutica abordada en este estudio, su enfoque se centra en el cliente, buscando comprenderlo mejor mediante la explotación de los datos disponibles. Nos encargamos de procesar y analizar datos, proponiendo una segmentación RFM a través del método no supervisado PAM y dicho enfoque permitió agrupar un conjunto específico de clientes objetivo, ofreciendo una comprensión más profunda de ellos. Se ha usado el método no supervisado PAM que sigue un proceso iterativo de partición similar al algoritmo k-means. Sin embargo, a diferencia de este último, PAM utiliza la mediana en lugar de la media como punto central en el proceso (centroide) debido a que la media es más sensible a valores atípicos en comparación de la mediana. El análisis y procesamiento de datos se realizó utilizando el lenguaje Python en Jupyter identificando 6 segmentos de clientes: Posible fidelizado, Satisfecho, Maduración, Desganado, Nuevo cliente y Potencial inactivo. Estos segmentos proporcionarán a la empresa una comprensión del comportamiento transaccional de sus clientes que les servirá de insumo en futuras tomas de decisiones
Abstract Currently, companies are faced with the challenge of managing significant volumes of data, optimizing its use and converting it into highly valuable information that generates brilliant and relevant ideas for decision-making and devising beneficial strategies. Across various disciplines, a commonly used technique is cluster analysis; however, its most beneficial application is observed in the business field, addressing aspects such as customer segmentation or classification and the detection of trends in purchasing patterns. In this way, the case of the virtual pharmaceutical store addressed in this study, its focus on the customer, seeking to better understand them by exploiting the available data. We took on the work of data processing and analysis, proposing an RFM segmentation using the unsupervised PAM method and this approach allowed us to group a specific set of target customers, offering a deeper understanding of them. The unsupervised PAM method has been used, which follows an iterative partitioning process similar to the k-means algorithm. However, unlike the latter, PAM uses the median instead of the mean as the center point in the process (centroid) because the mean is more sensitive to outliers compared to the median. Data analysis and processing was carried out using the Python language in Jupyter, identifying 6 customer segments: Possible loyal, Satisfied, Maturation, Reluctant, New customer and Potential inactive. These segments will provide the company with an understanding of the transactional behavior of its customers that will serve as input in future decision-making
Abstract Currently, companies are faced with the challenge of managing significant volumes of data, optimizing its use and converting it into highly valuable information that generates brilliant and relevant ideas for decision-making and devising beneficial strategies. Across various disciplines, a commonly used technique is cluster analysis; however, its most beneficial application is observed in the business field, addressing aspects such as customer segmentation or classification and the detection of trends in purchasing patterns. In this way, the case of the virtual pharmaceutical store addressed in this study, its focus on the customer, seeking to better understand them by exploiting the available data. We took on the work of data processing and analysis, proposing an RFM segmentation using the unsupervised PAM method and this approach allowed us to group a specific set of target customers, offering a deeper understanding of them. The unsupervised PAM method has been used, which follows an iterative partitioning process similar to the k-means algorithm. However, unlike the latter, PAM uses the median instead of the mean as the center point in the process (centroid) because the mean is more sensitive to outliers compared to the median. Data analysis and processing was carried out using the Python language in Jupyter, identifying 6 customer segments: Possible loyal, Satisfied, Maturation, Reluctant, New customer and Potential inactive. These segments will provide the company with an understanding of the transactional behavior of its customers that will serve as input in future decision-making
Description
Keywords
segmentación, clúster, métodos no supervisados, PAM