---

The aim of this project is to find a suitable forecasting model for the future value of reserve money of Bangladesh bank, reserve money of Bangladesh bank during the period 1991-92 to 2008-09 collected from statistics department of Bangladesh bank. Here, I use four statistical forecasting models: The Simple linear regression model, Log linear regression model, Holt´s linear exponential smoothing model and ARIMA. A comparison among four models reveals that Holt´s linear exponential smoothing model give less forecasting error than that others. So we proposed that for forecasting the future value of reserve money of Bangladesh bank, one can use the Holt´s linear exponential smoothing model. Here CHAPTER 1 contains introduction, background, objectives, limitations. CHAPTER 2 contains description of data. In CHAPTER 3 the methodology of the Simple linear regression model, Log linear regression model, Holt´s linear exponential smoothing method and ARIMA are described. Different model selection criteria and various measures of forecast error are also discussed here. CHAPTER 4 contains the analysis of the data. The CHAPTER 5 includes conclusion and finding.

---

Shamsul Alam, M.S: Studied Applied Statistics at University of Dhaka. Dhaka, Bangladesh.

---