Predicting market segmentation variables using Twitter following relations
From the beginning, social sciences have looked to categorize people into groups that share common characteristics, to better serve the population, giving a distinguished treatment to each group. Applying this approach to the planning of business activities, we can better understand people’s needs,...
| Autor Principal: | Brossard Núñez, Ian Paul |
|---|---|
| Formato: | info:eu-repo/semantics/masterThesis |
| Idioma: | Inglés |
| Publicado: |
Pontificia Universidad Católica del Perú
2018
|
| Materias: | |
| Acceso en línea: |
http://tesis.pucp.edu.pe/repositorio/handle/123456789/13072 |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| Sumario: |
From the beginning, social sciences have looked to categorize people into groups that share common characteristics, to better serve the population, giving a distinguished treatment to each group. Applying this approach to the planning of business activities, we can better understand people’s needs, choosing the most favorable marketing strategies for each stratum of customers (saving effort in advertising and distribution) and maximize the level of satisfaction of each of market segment. Social Media is not a stranger to this principle: a correct segmentation will allow companies to avoid bringing content to people that are not part of their target audience, and to better respond to comments and complaints about their products and brands. However, some Social Media like Twitter still haven’t included demographic markers about their users within their marketing platforms, rendering decision-making difficult. In this paper, we demonstrate that it is possible to estimate important demographic information in Social Media by analyzing the tastes and preferences of the users (represented through the Twitter accounts they follow). We present four predictive models that allowed us to estimate the gender, age, socio-economic level and LATIR Lifestyle of a Twitter user. These models were trained using machine learning algorithms |
|---|