Solving resource forecasting in WiFi Network by NeuralProphet

Time series forecasting needs several approaches such as data pretreatment, model construction, etc. During the covid 19 outbreak, the data is very dynamic, therefore data processing and appropriate modeling are worried. Identifying patterns, recognizing abnormal data points, is one of the first stages to enhancing forecast outcomes. A point is considered an anomalous point when it is far distant from the mean of the data series. In this research, we deploy an automated anomaly detection approach that incorporates data preparation of neuralprophet library. After that, we design a model via neuralprophet to predict data after preprocessing data. The strategy is evaluated on a dataset of the times that public wifi was used every day with the purpose of forecasting the value of the following 30 days. The anticipated outcome is compared with that of Prophet, hybrid AR-LSTM, consequently indicating that the suggested technique in the study offers the best outcomes.