Time Series Study on Household Final Consumption Expenditure in UK Using PCA and ARIMA Modeling

摘要

The main purpose of current research is to study the spending patterns of consumer and understand the change in consumer consumption expenditure during time. To achieve this goal, the suitable forecasting time series model is built using time series statistical data related to the final consumption expenditure (HFCE) incurred by households in United Kingdom (UK). Consumption data related to goods includes food and drink, alcohol and tobacco, clothing and footwear, household goods and services. Furthermore, services' consumption data contains housing, health, transport, communication, recreation and culture, education, restaurants and hotels, and other miscellaneous. In this study Principal Component Analysis (PCA) is conducted to reduce the high dimensionality of the independent variables/factors without the loss of much information into two principal components that accounts for as much variability in the data as possible. These two components basically reflect the consumer consumption index for primary needs and secondary needs. Different time series models on the consumer consumption index are investigated using the Box-Jenkins methodology that applies the Autoregressive Integrated Moving Average (ARIMA) models. The different models are evaluated and compared to find the best model that fits the consumer consumption indexes and finding on the most influencing factors are summarized.