Soiling in solar panel is unavoidable process which can cause major losses in solar panels. Estimation and mitigation of losses due to soiling is done through various methods. Comparison between three distinctive strategies for estimation of the impact of soiling on small scale photovoltaic plants is presented in this paper. A neural arrange show is utilized to appraise dirtying over a specific module utilizing yield control, light, and temperature as inputs. A comparison utilizing the different training-to-validation proportion is carried to choose the leading proportion for the show. There are customary strategies proposed to deterministically assess the solar power loss due to soiling. Due to intermittent nature of solar power, the error cannot be permanently eliminated, Hence, this paper proposes a statistical evaluation strategy ARIMA and neural network strategy ARIMA and LSTM to appraise the power loss by soiling on solar PV board. The test result based on real time dataset created indicates that LSTM is performing well even for data available for one month moreover RMSE value and predicted power loss in LSTM is more accurate.