Streamer is usually present at the initial stage of atmospheric pressure air discharge, which occurs in nature as a precursor to lightning, transient luminous events in upper atmosphere and has much potential applications in industry, such as the treatment of polluted gases/liquids, assisted combustion, plasma enhanced deposition etc. Streamer is a multi-scale problem both in time and in space, which brings much difficulty to the conventional diagnostic approaches. In past decades, fluid or particle-fluid hybrid models have been frequently used for understanding the mechanisms of streamer discharges because of their high efficiencies of calculations. Accuracies of the electron transport coefficients (including drift/diffusion coefficient, ionization/attachment coefficient, electron mean energy and extra) play a key role in ensuring the correctness of the fluid or hybrid simulations. As far as we know, BOLSIG+ and MAGBOLTZ are two typical tools for obtaining the electron transport coefficients and have been widely utilized in previous models. BOLSIG+ uses two-term approximation which is not sufficient for some molecular gases, MAGBOLTZ cannot calculate the bulk transport coefficients:these data are required for some models. METHES is an additional tool for computing electron transport coefficients, however, specific platform is required which is not very user-friendly. As sorts of drawbacks exist in currently available calculating tools, in the paper, a Monte Carlo model is developed for computing the electron transport coefficients in gases, the model is flexible to choose any type of gas mixture and its accuracy has been validated by comparing with BOLSIG+ and METHES. Furthermore, the influences of N2-O2 mixture and three-body attachment process in high gas pressures on the transport coefficient are investigated. It is worth mentioning that three-body attachment process can significantly change the electron transport properties at a relatively low reduced electric field. Thus, specific attention must be paid to the transport coefficients if simulation is performed at a high pressure. In addition, differences between the bulk and flux coefficients are analyzed which are not distinguished in some previous models. Finally, we further validate the present Monte Carlo model by performing simulation of streamer discharge in atmospheric N2, which shows that the improved electron transport coefficient from our Monte Carlo model can improve the simulated plasma properties, in particular at the interior of the streamer channel. The existence of divergence at the tip of the streamer channel might be due to our local field approximation; if a density gradient term is included in the impact ionization term and local electron energy approximation of the electron transport coefficients is used, the accuracy of the fluid can be improved further.