Mobile adhoc networks are composed of anticipating nodes which forward data to each other. They are organised in temporary topology. In such a network Quality of service is of prime importance. Delay is essential metric of QoS. E2E delay depends on many important variables such as path length, hop count, mobility of nodes, intervention from other neighbors, bandwidth and past values of delay etc. In order to make data delivery more reliable, accurate prediction of delay is necessary task. This paper is directed to analyse the delay of packet delivery in adhoc networks on regression graph. In this paper we have tried to compare the predictions done by time series prediction model of ANN for three routing schemes. We have simulated AODV, DSDV and OLSR on NS3 to collect the data and feed forward neural networks are used to predict the delay.