Comparative Performance of Prediction Methods for Digital Wallet Transactions in the Pandemic Period

Abstract

A pandemic situation such as Covid-19 which is still ongoing has given significant impacts to various sectors such as education, economy, tourism, and social which is in turn impacting the community at a national scale. On the other hand, the pandemic situation has also brought a positive impact on companies engaged in finance that utilizes information technology, namely digital wallets, a company that runs a market place in the digital world. In an effort to anticipate a dynamic market place, the company needs to predict the movement of transactions from time to time by building a model and performain the simulation to such model. Based on this problem, this paper presents simulations on the prediction models based on methods namely, naïve, Single Moving Average (SMA), Exponential Moving Average (EMA), combined SMA-naive methods, combined EMA-naive methods, as well as did the comparison of the best performance of every model by using Mean Absolute Percentage Error (MAPE) measurement. From the results of comparison, it is concluded that exponential moving average method delivers the best performance as prediction tool with MAPE of 23,4%.

Manufacturing a Plotter Printer with Computer Numerical Control Based Pen Ink Using CoreXY Mechanisms

“…Mean absolute percentage error (MAPE) has been employed in many machine learning and regression precision analysis [3]. One of the earliest discussions on plotter printing can be found in [4].…”