Inflation forecasting using hybrid ARIMA-LSTM model

Prediction of time series is one of the most demanding research areas due to the nature of various
\ntime series i.e., stocks, inflation, stock indexes etc. Various methods have been used in the past to
\nforecast such time series, however, Machine Learning (ML) methods have been suggested in the
\nacademic literature as alternatives to statistical ones for time series forecasting. Yet, scant
\nevidences are available about their relative performance in order of their accuracies and
\ncomputational requirements. In this thesis, a hybrid model consisting of ARIMA and LSTM is
\nproposed and compared with individual models ARIMA, LSTM, and PROPHET for inflation
\nforecasting. Two Scale-dependent metrics namely mean absolute error (MAE) and root mean
\nsquare error (RMSE), one Percentage-error metric, mean absolute percentage error (MAPE) and
\ncoefficient of determination (R
\n2
\n) are used to evaluate the variance between dependent and
\nindependent parameters for inflation forecasting in developed and developing countries.
\nConsumer Price Index (CPI) data is collected monthly to reflect the effect of price inflation at
\nconsumer level. Most of the central banks depend on inflation forecast to inform their respective
\nmonetary policy makers and to enhance the efficacy of monetary policy. The publicly available
\nCPI data is presented for analysis and evaluation of price inflation effects on developed and
\ndeveloping countries. For this research work, six developed countries (Canada, United States,
\nAustralia, Norway, Poland and Switzerland) and six developing countries (Colombia, Indonesia,
\nBrazil, South Africa, India and Mexico) with different durations are targeted to evaluate the
\nperformances of proposed machine learning model and the individual models to forecast inflation
\n(CPI). The proposed HYBRID model with one-step ahead forecasting outperformed every other
\nmodel for forecasting inflation (CPI) of developed and developing countries regardless of duration.
\nThe best performance was observed by taking 90% training data and 10% testing data. All
\niv
\nforecasting models performed better on data of six developed countries with overall average errors
\nof 1.023796 in MAE, 0.009648 in MAPE and 1.222454 in RMSE when taking 10% as test data.
\nWhile in the case of developing countries overall average errors of MAE, MAPE and RMSE was
\n1.361308, 0.011847, and 1.562288 respectively. Also, in the case of 20% and 30% test data, the
\nperformance of all models on developed countries data was better than developing countries in
\nterms of least errors in MAE, MAPE and RMSE.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.