atural gas(NG) is an important energy resources that is becoming more and more popular because of its environmental benefits(lower impact on environmental pollution). All most all developed countries are concerned about the natural gas consumption due to low reserve. Omer Fahrettin Dem, IREL, Selim Zaim, 2011 predicted that china's natural gas consumption will continue to grow and expected to achieve 354.1bcm by 2020. Michael Ratner(India), 2017 on his research article, he found that the natural gas portion in india's energy mix is 7% as it remains small compared to that of the US and other developed countries like Brazil, China. India's target it to double the proportion of natural gas consumption by 2022. To achieve this goal we would require major upstream, midstream and downstream investments as well as the continued political will to take necessary steps and to decrease reliance on coal and oil. Therefore, demand for this source of energy has in creased considerably in recent years. The largest increase in world's primary energy consumption is attributed to N.G as per U.S Energy Information Administration 2016.

It is projected that the Natural Gas consumption as primary energy source will increase to 2040 TCF compared to the recorded consumption 120 TCF in the year 2012-2013. As per British Petroleum(BP) global 2015 NG contributes 23.8% of the primary energy consumption globally and remains as the main fuel in production of electricity and as a fuel for the industry. monthly, quarterly and yearly basis. It has reviewed that the computational models were suitable for natural gas consumption for a better input parameters. The model efficiency not only depends upon the algorithm but majorly depends upon the input parameters. Every natural gas distributor is obliged to make a nomination of natural gas by its supplier, which is the amount of gas needed for the future days. There is a certain regulated tolerance that is allowed. In case the actual consumption exceeds the nominated amount, the distributor must pay a certain penalty. On the other hand, if nominated amount exceed actual consumption, different type of penalty will be charged as well. Since the incorrect nominations lead to high costs, accurate predictions of natural gas consumption for the following day are very important due to financial reasons.

For the research purposes, literature overview analysis was conducted using PRQUEST database. The keywords "natural gas consumption", " "prediction of Natural Gas" OR "demand of Natural Gas", "Consumption of Natural Gas", "Prediction models in Natural Gas)", were used for searching articles. The articles were searched within three indexes: Science Citation Index Expanded (SCI-Expanded), Social Science Citation Index (SCI), and Arts and Humanities Citation Index (A&HCI) for the period of 2002 to 2017.

This search resulted with 276 papers, including article (201), proceedings paper (28) and review (47). After reviewing the title, abstracts and keywords of all found articles, articles that are not related to models for prediction of CNG for residential or commercial use were eliminated. Thereafter, 72 articles remain that met posted criteria. Those papers were analysed according to several criteria: methods used for predictions of CNG, input variables used for modelling, prediction area and prediction horizon.

Similar literature review was conducted by Soldo (2012), who analysed natural gas consumption from the year 1949 to 2010 and Dario Sebalj, Josip Mesarie, Davor Dujak , 2017, he predicted the Natural Gas consumption from the year 2002-2017 using web of science core collection (WOSCC) database. Year 2018 J As it can be seen in Table 1, in the last three years 34 papers considering natural gas prediction were A number of researchers attempted to develop models for the prediction of NGC on daily, weekly, published, which is more than 55% of all analysed papers. 2016),who fore casted NG consumption using time series methods.

There are four papers in which weekly prediction of NG consumption was reported. Those are papers writ in by potocnik at al. ( 2007), who proposed a forecasting model in order to fore cast risk estimation, and Kaynar et al. (2011), who used neural network and neuro fuzzy system for prediction of NG consumption on weekly basis. Dejan Ivezi? ( 2006) predicts natural gas consumption on weekly basis by using ANN model in Belgrade, Serbia. Ma?gorzata Trojanowska (2014) also predict on weekly basis by using Regression model in Poland.

There are many number of authors predicted NG consumption on daily level.

Gil & Deferrari ( 2004) proposed a daily prediction model in Argentina, Steven R. Vitullo, Ronald H. Brown; George F. Corliss, Brian M. Marx.( 2009 Tonkovic at al.), who created a prediction model of NG consumption by using neural networks on a regional level on hourly scale for predicting NG consumption, Sabo etal. ( 2011), who proposed mathematical models of natural gas consumption, and Szoplik (2015), who forecasted NG consumption in Poland using neural network models. Krzysztof N?cka, Ma?gorzata & Trojanowska who created a prediction model by using Regression model on a particular area on hourly basis to predict Natural Gas consumption.

There are only seven papers that predicted natural gas consumption on regional level. Gil & Deferrari (2004) presented there sults for the case of Greater Buenos Aires region in Argentina. DejanIvezi?. ( 2006) investigated the prediction of NG consumption in the region of Belgrade, Serbia to predict the Natural Gas consumption using Parameters of ANN are obtained from the historical data using a Levenberg-Marquardt training algorithm.

Nil Aras ( 2008), Beyzanur Cayir Ervural Omer Faruk Beyca Selim Zaim(2016) used genetic algorithm to predict NGC of Turkey city Eskisehir. Istanbul, Omer Fahrettin DEM ? IREL, Selim ZAIM, proposed neural networks and multivariate time series models to predict Natural Gas consumption for the city of Istanbul. Ahmet Goncu Mehmet, Oguz Karahan & Tolga Umut Kuzuba¸ (2013) propose a methodology which combines natural gas demand estimation with a stochastic temperature model. The model demand and temperature processes separately and derive the distribution of natural gas consumption with a conditional temperature. Hossein Iranmanesh, Majid Abdollahzade & Arash Miranian (2011) predict natural gas consumption using PSO (2017). This model hybrid computational intelligence model was tested for its robustness by prediction of day ahead natural gas demands.

Another hybrid model consisting of ANFIS and computer simulation was proposed by Azadeh et al. (2015) have presented ANFIS based techniques. The ANFIS is was also used by Kaynar et al. (2011) to predict weekly NG consumption in turkey.

Ma & Li (2010) predicted NG consumption based on the Grey system model. The same approach was using Boran ( 2015

The study was conducted with seventy two number of research articles of Natural Gas prediction for various Countries in different mathematical and scientific methods. It was found that China and other developed countries are focusing this type of study for weekly, monthly and yearly basis. ANN is the most appropriate techniques for the prediction of Natural Gas consumption. The different researchers applied genetic algorithm, feed forward, Back propagation & PSO methods for this prediction. All most all researchers are agreed upon the other popular methods are neuro-fuzzy inference system, genetic algorithms, time series methods,, support vector machines/ regression, Grey system models, mathematical and statistical models orhybrid models based on several methods. Some researches use two or more methods in the same paper. But analysis has shown that for modeling, authors often use past NG consumption data and weather data (mostly temp.)as input variables. Other variables include month, days of the week, wind speed, temperature, humidity & price number of natural gas subscribers, GDP, inflation rate etc. Speaking of prediction are as, it can be seen that most of the papers deal with the predictions on country level. Predictions can be made as well as on regional, city, or even house level.