Inflation expectations and macroeconomic data from Brazil

This dataset shows consumer inflation expectations and raw macroeconomic data from Brazil. Using this data source it was performed an attribute selection on the macroeconomic data to find the best variables to forecast the 12-month future inflation of Brazil (dependent variable). The dataset also shows what were the forecasts for an interval of 20 months from the following models: AR(1), multiple linear regression, an LSTM neural network with macroeconomic data and the same neural network with the addition of inflation expectations.

本数据集收录了巴西消费者通胀预期与原始宏观经济数据。本研究依托该数据源对宏观经济数据开展属性筛选，以遴选出最优变量用于预测巴西12个月远期通胀（因变量）。此外，本数据集还提供了以下模型针对20个月预测区间的通胀预测结果：AR(1)模型、多元线性回归模型、仅使用宏观经济数据的长短期记忆（LSTM）神经网络，以及加入通胀预期变量的同款LSTM神经网络。