Cite Article

The Consumer Price Index (CPI) is one of the most important economic indicators for countries’ characterization and is typically considered an official measure of inflation. The CPI considers the monthly price variation of a determined set of goods and services in a specific region, and it is key in the economic and social planning of a given country, hence the great importance of CPI forecasting. In this paper, we outline a comparative study of state-of-the-art predictive models over an Ecuadorian CPI dataset with 174 monthly registers, from 2005 to 2019. This small available dataset makes forecasting a challenging time-series-prediction task. Another difficulty is last year´s trend variation, which since mid-2016, has changed from an upward average of 3.5 points to a stable trend of ±0.8 points. This paper explores the performance of relevant predictive models when tackling the Ecuadorian CPI forecasting problem accurately for the next 12 months. For this, a comparative study considering a variety of predictive models is carried out, including the Neural networks approach using a Sequential Model with Long Short-Term Memory layers machine learning using Support Vector Regression, as well as classical approaches like SARIMA and Exponential Smoothing. We also consider big corporations´ tools like Facebook Prophet. As a result, the paper presents the best predictive models, and parameters found, along with Ecuador´s CPI forecasting for the next 12 months (part of 2020). This information could be used for decision-making in several important topics related to social and economic activities.

The Consumer Price Index (CPI) is one of the most important economic indicators for countries’ characterization and is typically considered an official measure of inflation. The CPI considers the monthly price variation of a determined set of goods and services in a specific region, and it is key in the economic and social planning of a given country, hence the great importance of CPI forecasting. In this paper, we outline a comparative study of state-of-the-art predictive models over an Ecuadorian CPI dataset with 174 monthly registers, from 2005 to 2019. This small available dataset makes forecasting a challenging time-series-prediction task. Another difficulty is last year´s trend variation, which since mid-2016, has changed from an upward average of 3.5 points to a stable trend of ±0.8 points. This paper explores the performance of relevant predictive models when tackling the Ecuadorian CPI forecasting problem accurately for the next 12 months. For this, a comparative study considering a variety of predictive models is carried out, including the Neural networks approach using a Sequential Model with Long Short-Term Memory layers machine learning using Support Vector Regression, as well as classical approaches like SARIMA and Exponential Smoothing. We also consider big corporations´ tools like Facebook Prophet. As a result, the paper presents the best predictive models, and parameters found, along with Ecuador´s CPI forecasting for the next 12 months (part of 2020). This information could be used for decision-making in several important topics related to social and economic activities.

The Consumer Price Index (CPI) is one of the most important economic indicators for countries’ characterization and is typically considered an official measure of inflation. The CPI considers the monthly price variation of a determined set of goods and services in a specific region, and it is key in the economic and social planning of a given country, hence the great importance of CPI forecasting. In this paper, we outline a comparative study of state-of-the-art predictive models over an Ecuadorian CPI dataset with 174 monthly registers, from 2005 to 2019. This small available dataset makes forecasting a challenging time-series-prediction task. Another difficulty is last year´s trend variation, which since mid-2016, has changed from an upward average of 3.5 points to a stable trend of ±0.8 points. This paper explores the performance of relevant predictive models when tackling the Ecuadorian CPI forecasting problem accurately for the next 12 months. For this, a comparative study considering a variety of predictive models is carried out, including the Neural networks approach using a Sequential Model with Long Short-Term Memory layers machine learning using Support Vector Regression, as well as classical approaches like SARIMA and Exponential Smoothing. We also consider big corporations´ tools like Facebook Prophet. As a result, the paper presents the best predictive models, and parameters found, along with Ecuador´s CPI forecasting for the next 12 months (part of 2020). This information could be used for decision-making in several important topics related to social and economic activities.