scholarly journals Long-Term Electricity Demand Prediction via Socioeconomic Factors—A Machine Learning Approach with Florida as a Case Study

Energies ◽  
2020 ◽  
Vol 13 (15) ◽  
pp. 3996 ◽  
Author(s):  
Marwen Elkamel ◽  
Lily Schleider ◽  
Eduardo L. Pasiliao ◽  
Ali Diabat ◽  
Qipeng P. Zheng
Keyword(s):  
Time Series ◽  
Regression Model ◽  
Multiple Regression ◽  
Energy Demand ◽  
Mean Squared Error ◽  
Multiple Regression Model ◽  
Electricity Demand ◽  
Degree Days ◽  
Population Number ◽  
Demand Prediction

Predicting future energy demand will allow for better planning and operation of electricity providers. Suppliers will have an idea of what they need to prepare for, thereby preventing over and under-production. This can save money and make the energy industry more efficient. We applied a multiple regression model and three Convolutional Neural Networks (CNNs) in order to predict Florida’s future electricity use. The multiple regression model was a time series model that included all the variables and employed a regression equation. The univariant CNN only accounts for the energy consumption variable. The multichannel network takes into account all the time series variables. The multihead network created a CNN model for each of the variables and then combined them through concatenation. For all of the models, the dataset was split up into training and testing data so the predictions could be compared to the actual values in order to avoid overfitting and to provide an unbiased estimate of model accuracy. Historical data from January 2010 to December 2017 were used. The results for the multiple regression model concluded that the variables month, Cooling Degree Days, Heating Degree Days and GDP were significant in predicting future electricity demand. Other multiple regression models were formulated that utilized other variables that were correlated to the variables in the best-selected model. These variables included: number of visitors to the state, population, number of consumers and number of households. For the CNNs, the univariant predictions had more diverse and higher Root Mean Squared Error (RMSE) values compared to the multichannel and multihead network. The multichannel network performed the best out of the three CNNs. In summary, the multichannel model was found to be the best at predicting future electricity demand out of all the models considered, including the regression model based on the datasets employed.

FORECASTING CONSUMER PRICE INDEX (CPI) USING TIME SERIES MODELS AND MULTI REGRESSION MODELS (ALBANIA CASE STUDY)

2018 ◽  
Author(s):  
Eralda Gjika Dhamo ◽  
Llukan Puka ◽  
Oriana Zaçaj
Keyword(s):  
Time Series ◽  
Regression Model ◽  
Multiple Regression ◽  
Consumer Price Index ◽  
Multiple Regression Model ◽  
Time Series Models ◽  
Financial Indicators ◽  
Consumer Prices ◽  
Approach Time

In this work we analyse the CPI index as the official index to measure inflation in Albania, Harmo-nized Indices of Consumer Prices (HICPs) as the bases for comparative measurement of inflation in European countries and other financial indicators that may affect CPI. This study is an attempt to model CPI based on combination of multiple regression model with time series forecasting models. In the first approach, time series models were used directly on the CPI time series index to obtain the forecast. In the second approach, the time series models (SARIMA, ETS) were used to model and simulate forecast for each subcomponent with significant correlation to CPI and then use the multiple regression model to obtain CPI forecast. The projection of this indicator is important for understand-ing the country's economic and social development. This study helps researchers in the field of time series modeling, economic analysis and investments.

scholarly journals Evaluation of a Multiple Regression Model for Noisy and Missing Data

2018 ◽  
Vol 8 (4) ◽  
pp. 2220
Author(s):  
Chanintorn Jittawiriyanukoon
Keyword(s):  
Big Data ◽  
Missing Data ◽  
Data Collection ◽  
Regression Model ◽  
Multiple Regression ◽  
Mean Squared Error ◽  
Real Data ◽  
Multiple Regression Model ◽  
Average Percentage ◽  
Standard Data

<span>The standard data collection problems may involve noiseless data while on the other hand large organizations commonly experience noisy and missing data, probably concerning data collected from individuals. As noisy and missing data will be significantly worrisome for occasions of the vast data collection then the investigation of different filtering techniques for big data environment would be remarkable. A multiple regression model where big data is employed for experimenting will be presented. Approximation for datasets with noisy and missing data is also proposed. The statistical root mean squared error (RMSE) associated with correlation coefficient (COEF) will be analyzed to prove the accuracy of estimators. Finally, results predicted by massive online analysis (MOA) will be compared to those real data collected from the following different time. These theoretical predictions with noisy and missing data estimation by simulation, revealing consistency with the real data are illustrated. Deletion mechanism (DEL) outperforms with the lowest average percentage of error.</span>

scholarly journals Multiple Regression Model Fitted for Rice Production Forecasting in Nepal: A Case of Time Series Data

2018 ◽  
Vol 2 ◽  
pp. 89-98
Author(s):  
Chuda Prasad Dhakal
Keyword(s):  
Time Series ◽  
Regression Model ◽  
Multiple Regression ◽  
Rural Population ◽  
Time Series Data ◽  
Multiple Regression Model ◽  
Rice Production ◽  
Series Data ◽  
Agricultural Labor ◽  
Production Forecasting

Background: Fitting a multiple regression model is always challenging and the level of difficulty varies according to the purpose for which it is fitted. Two major difficulties that arise while fitting a multiple regression model for forecasting are selecting 'potential predictors' from numerous possible variables to influence on the forecast variable and investigating the most appropriate model with a subset of the potential predictors.Objective: Purpose of this paper is to demonstrate a procedure adopted while fitting multiple regression model (with an attempt to optimize) for rice production forecasting in Nepal.Materials and Methods: This study has used fifty years (1961-2010) of time series data. A list of twenty-one predictors thought to impact on rice production was scanned based upon past literature, expert's hunches, availability of the data and the researcher's insight which left eleven possible predictors. Further, these possible predictors were subjected to family of automated stepwise methods which left five ‘potential predictors’ namely harvested area, rural population, farm harvest price, male agricultural labor force and, female agricultural labor force. Afterwards, best subset regression was performed in Minitab Version 16 which finally left three 'appropriate predictors' that best fit the model namely harvested area, rural population and farm harvest price.Results: The model fit was significant with p < .001. Also, all the three predictors were found highly significant with p < 0.001. The model was parsimonious which explained 93% variation in rice production with 54% overlapping predictive work done. Forecast error was less than 5%.Conclusion: Multiple regression model can be used in rice production forecasting in the country for the enhanced ease and efficiency.Nepalese Journal of Statistics, Vol. 2, 89-98

Extended SPRT for structural change detection of time series based on a multiple regression model

2010 ◽  
Vol 15 (4) ◽  
pp. 417-420 ◽  
Author(s):  
Katsunori Takeda ◽  
Tetsuo Hattori ◽  
Tetsuya Izumi ◽  
Hiromichi Kawano
Keyword(s):  
Time Series ◽  
Structural Change ◽  
Change Detection ◽  
Regression Model ◽  
Multiple Regression ◽  
Multiple Regression Model

Factors Affecting NPAs in Indian Banking Sector

Paradigm ◽  
2021 ◽  
Vol 25 (2) ◽  
pp. 181-193
Author(s):  
Nitya Garg
Keyword(s):  
Regression Model ◽  
Multiple Regression ◽  
Banking Sector ◽  
Relative Weight ◽  
Multiple Regression Model ◽  
Factors Affecting ◽  
Weight Analysis ◽  
Specific Factors ◽  
Relative Weight Analysis ◽  
Bank Specific Factors

Banking sector is the backbone of any economy, so it is necessary to focus on its performance which is largely affected by its non-performing assets (NPAs). In the year 2018–2019, NPA of scheduled banks was Rs 355,076 Crore which is 3.7% of net advances. The purpose of this study is to identify the determinants based on analysis from previous literatures, and majorly macroeconomic and bank specific factors which are affecting NPAs using the relative weight analysis and to frame a model to predict future NPAs using multiple regression model using SPSS. The study also attempts to focus on actions and remedies that banks should make to control future NPAs. Findings of the study will act as a scaffolding for financial analysts and policymakers to prevent the conversion of its performing assets into NPAs and also help in proper management of banks and also in the recovery of economy.

Sign in / Sign up

Export Citation Format

Share Document