Advanced

Using Conjunctural Indices in Prediction Models for Gas Sales---A Case Study

Sigot, Adrian (2010) In MSc Theses
Department of Automatic Control
Abstract
This thesis evaluates whether it is possible to use conjunctural indices from Statistics Sweden (Statistiska Centralbyrån) in order to improve the sales predictions at a Swedish gas company. For this purpose sales data for three of the main gases are gathered from the year 1995 and forth. In the first step, this data is used to create ARX and State Space models. Sales predictions are made from these models and the quality of these predictions is measured. In the second step, the conjunctural indices from Statistics Sweden for the corresponding period is used as input values and new ARX and State Space models are created. Predictions are made also with these models and are then compared to the predictions without help from the conjunctural... (More)
This thesis evaluates whether it is possible to use conjunctural indices from Statistics Sweden (Statistiska Centralbyrån) in order to improve the sales predictions at a Swedish gas company. For this purpose sales data for three of the main gases are gathered from the year 1995 and forth. In the first step, this data is used to create ARX and State Space models. Sales predictions are made from these models and the quality of these predictions is measured. In the second step, the conjunctural indices from Statistics Sweden for the corresponding period is used as input values and new ARX and State Space models are created. Predictions are made also with these models and are then compared to the predictions without help from the conjunctural indices. The results are however not significantly better. Correlation analysis show strong correlation between the conjunctural indices and the sales for some intervals in the available sales period, but no thoroughgoing correlation and no strong correlation at the current time (year 2008-2010). The conclusion is thus that the available conjunctural indices cannot be used to improve the sales predictions. However, since the correlations have been strong during come periods, it could be worth monitoring the correlations in case they reappear (Less)
Please use this url to cite or link to this publication:
author
Sigot, Adrian
supervisor
organization
year
type
H3 - Professional qualifications (4 Years - )
subject
publication/series
MSc Theses
report number
TFRT-5862
ISSN
0280-5316
language
English
id
8847471
date added to LUP
2016-03-16 12:44:01
date last changed
2016-03-16 12:44:01
@misc{8847471,
  abstract     = {This thesis evaluates whether it is possible to use conjunctural indices from Statistics Sweden (Statistiska Centralbyrån) in order to improve the sales predictions at a Swedish gas company. For this purpose sales data for three of the main gases are gathered from the year 1995 and forth. In the first step, this data is used to create ARX and State Space models. Sales predictions are made from these models and the quality of these predictions is measured. In the second step, the conjunctural indices from Statistics Sweden for the corresponding period is used as input values and new ARX and State Space models are created. Predictions are made also with these models and are then compared to the predictions without help from the conjunctural indices. The results are however not significantly better. Correlation analysis show strong correlation between the conjunctural indices and the sales for some intervals in the available sales period, but no thoroughgoing correlation and no strong correlation at the current time (year 2008-2010). The conclusion is thus that the available conjunctural indices cannot be used to improve the sales predictions. However, since the correlations have been strong during come periods, it could be worth monitoring the correlations in case they reappear},
  author       = {Sigot, Adrian},
  issn         = {0280-5316},
  language     = {eng},
  note         = {Student Paper},
  series       = {MSc Theses},
  title        = {Using Conjunctural Indices in Prediction Models for Gas Sales---A Case Study},
  year         = {2010},
}