Yes, that is what is being done in short term load forecasting. Of course the prediction accuracy will depend on the length of data available and the stationary ( time invariant) statistical characteristics of the data used. 

Dear Rajeev Kumar Ji,

You know the numerical methods and analysis of extrapolations, that will show you the solution of next day of load forecasting but the problem it does need large amount data of previous years. Actually, how large information we need it depends on how much error of calculation we can have. More data will give you less error of forecasting data. May be tomorrow I am going to buy three giant Refrigerator and three big air conditions. This case will influence the power demand for next date in account. Ultimately we just make assumption of demand of electricity. On the other hand, we do really collect information of local area what classes people are they living. what sort of appliances they use and a lots of information is required. then we try to plot the data in sheets and do some mathematical calculations. I worked in Bangladesh in power system. Like developing country in Bangladesh, demand of electricity is much higher than we forecast because every moment people open new small and medium industry and home appliances. On the contrary, it is quiet easy to tell the load forecast in Japan because they don't have such fluctuation like in Bangladesh or like India. Japan has almost constant level of demand of electricity. Although the amount of electricity may change little bit due to environmental facts.   

Thank you so much

Hi Kumar,

We have used Artificial Neural Networks to predict the properties of HV insulation under thermal ageing until 5 thousands hours of ageing time) . In your case it's possible to use this method but you have to dispose of sufficient data. Take the statistics of last 10 years and study the variation of power demand for each month (January during the 10 years, and so for every month - because the demand is not the same in each season and all months) taking into consideration the human and industrial development factors. 

Dear Rajeev,

I fully agree to  Md. Ruhul Amin's explanation, there is a lot of parameters which decides the electricity requirement accurately, like  Md. Ruhul Amin points out in developed countries you may predict more accurately (with relatively less error) as there is established industries who correctly predicts how much electricity they require seeing the previous history, that is not true at all in case of developing countries where the demand of electricity varies a lot due to setting up industries year over year as the country is still in developing stage. Depends more on availability of the energy sources, more probable of absence of sources nearby.

 Thanks to all of you for giving nice suggestions.

In case of short term load forecasting, you can of course predict the next day load demand using a one month load demand history. Different tools can be used like neural network, genetic algorithm etc. But the extent of accuracy of  prediction depends on the the amount of data of the past years used for training the network. With you data you can still predict with estimated error of 2 - 4%.  You can check the file attachment for different methods to do the load forecasting. Good luck!