TWI888832B - Server managing change transfer based on time series model - Google Patents

Abstract

A server managing change transfer based on a time series model is provided. The server includes a storage medium, a transceiver and a processor. The processor establishes a time series model corresponding to a change type; the processor uses the time series model to obtain a target-date-sequence corresponding to the change type, wherein the target-date-sequence corresponds to a target-predicted-that-day-change-amount, and the target-predicted-that-day-change-amount is less than or equal to a change threshold value; the processor uses a predicted-date-number and a preset-advance-date-number to obtain a date-number-to-be-transferred-in-advance, and transmits a change transfer request to an external bank electronic device through the transceiver when date-number-to-be-transferred-in-advance, wherein the predicted-date-number is associated with the target-date-sequence.

Description

A server that manages small change inbound calls based on a time series model

The present disclosure relates to a server for managing change inflows based on a time series model.

At present, when bank customers need change, they usually exchange it with the bank. For banks, it is difficult to effectively manage the amount of change, resulting in the need to temporarily transfer change from external banks. How to effectively manage the transfer of change is one of the goals that technical personnel in this field should strive for.

The server disclosed herein for managing change transfer based on a time series model includes a storage medium, a transceiver, and a processor. The transceiver is communicatively connected to an external bank electronic device. The processor is coupled to the storage medium and the transceiver, wherein the processor establishes a time series model corresponding to the change type; the processor uses the time series model to obtain a target date sequence corresponding to the change type, wherein the target date sequence corresponds to a target predicted change amount for the day, and the target predicted change amount for the day is less than or equal to the change threshold; the processor uses the predicted date number and the preset advance date number to obtain the number of dates that need to be transferred in advance, and when the number of dates needs to be transferred in advance, the processor transmits a change transfer request to the external bank electronic device through the transceiver, wherein the predicted date number is related to the target date sequence.

FIG1 is a schematic diagram of a server 100 for managing change transfer based on a time series model according to an embodiment of the present disclosure. The server 100 may include a storage medium 110, a transceiver 120, and a processor 130. The transceiver 120 is communicatively connected to an external bank electronic device 200. The processor 130 is coupled to the storage medium 110 and the transceiver 120. In other embodiments, the server 100 may include an input/output device 140 coupled to the processor 130. The server 100 may be set in a bank.

The storage medium 110 may include any type of fixed or removable random access memory (RAM), read-only memory (ROM), flash memory, hard disk drive (HDD), solid state drive (SSD) or similar elements or a combination of the above elements, and is used to store multiple modules or various applications that can be executed by the processor 130.

The transceiver 120 transmits signals wirelessly or wiredly.

The processor 130 may include a central processing unit (CPU), or other programmable general-purpose or special-purpose micro control unit (MCU), microprocessor, digital signal processor (DSP), programmable controller, application specific integrated circuit (ASIC), graphics processing unit (GPU), image signal processor (ISP), image processing unit (IPU), arithmetic logic unit (ALU), complex programmable logic device (CPLD), field programmable gate array (FPGA), or other similar components or combinations of the above components. The processor 130 may access and execute multiple modules and various applications stored in the storage medium 110 .

The input/output device 140 may include various input/output devices such as a screen or a speaker.

In this embodiment, the storage medium 110 can store the date sequence corresponding to the change type, the amount of change on the training day corresponding to the change type, the amount of change on the day before the training corresponding to the change type, the amount of change exchanged in the past week corresponding to the change type, the amount of change deposited in the past week corresponding to the change type, the average amount of change in the past week corresponding to the change type, the average amount of change in the past month corresponding to the change type, and whether the change type is a call-in date sequence. Table 1 is an example of the data stored in the storage medium 110. Table 1 Change Type Date sequence Amount of change on the training day Amount of change before training Amount of change exchanged in the past week Amount of change deposited in the past week Average quantity in the past week Average quantity in the past month Is it the transfer date sequence? A 1 1200 0 50 0 650 700 1 (is the order of the transfer date) A 2 1150 1200 60 10 800 750 0 (non-transferred date sequence) … … … … … … … … … A 100 900 1000 20 120 700 780 0 (non-transferred date sequence)

Based on the data in Table 1, the processor 130 can establish a time series model corresponding to the change type. In detail, the time series model can be an ARIMA model, a SARIMA model, or an LSTM (Long Short-Term Memory) model. Furthermore, the variables of the time series model can include the date sequence in Table 1, the amount of change before training, the amount of change exchanged in the past week, the amount of change deposited in the past week, the average amount in the past week, the average amount in the past month, and whether it is a call-in date sequence.

It should be noted here that the processor 130 can obtain the required quantity of change to be transferred in based on the data in Table 1. Specifically, the date sequence may include a first date sequence, and the amount of change on the training day may include the amount of change on the first training day, wherein the first date sequence corresponds to the amount of change on the first training day, and the first date sequence corresponds to "is a transfer-in date sequence". For example, among the date sequences in Table 1, since the "Is it a transfer-in date sequence" corresponding to the date sequence "1" is "is a transfer-in date sequence", the date sequence "1" is the above-mentioned first date sequence, and then the processor 130 can set the first training day's change amount "1200" corresponding to the first date sequence "1" as the required quantity of change to be transferred in. The purpose of the required quantity of change to be transferred in will be further explained later.

After establishing a time series model corresponding to the change type, the processor 130 may use the time series model to obtain a target date sequence corresponding to the change type, wherein the target date sequence corresponds to a target predicted change amount for the day, and the target predicted change amount for the day is less than or equal to the change threshold. In more detail, the processor 130 may use the time series model to obtain multiple predicted change amounts for the day, and the multiple predicted change amounts for the day correspond to multiple date sequences, wherein the multiple predicted change amounts for the day include the target predicted change amount for the day, and the multiple date sequences include the target date sequence. This will be further described below.

FIG. 2 is a schematic diagram of a curve 22 representing multiple predicted daily change amounts according to an embodiment of the present disclosure. Please refer to FIG. 1 and FIG. 2 at the same time. It should be noted that the date sequence and the (training) daily change amount shown in Table 1 above correspond to the curve 21 of FIG. 2, and as shown in the curve 21, the daily change amount will be less than or equal to the change threshold value on the date sequence "200th day". In other words, the change transfer was executed on the date sequence "200th day".

It is assumed that the processor 130 obtains the curve 22 using the time series model, wherein the curve 22 includes points 22-1, 22-2, 22-3, ... until point 22-n. Specifically, point 22-1 represents the predicted amount of change for the day y ₁ corresponding to date sequence x ₁ , point 22-2 represents the predicted amount of change for the day y ₂ corresponding to date sequence x ₂ , point 22-3 represents the predicted amount of change for the day y ₃ corresponding to date sequence x ₃ , ... until point 22-n represents the predicted amount of change for the day y _n corresponding to date sequence x _n . Since the predicted amount of change for the day y _n is less than or equal to the change threshold, the processor 130 can obtain the target date sequence as date sequence x _n . In other words, the processor 130 can obtain (predict) that the target predicted daily change quantity at the "300th day" of the target date sequence x _n will be less than or equal to the change threshold. In other words, change transfer needs to be executed again at the "300th day" of the target date sequence. Furthermore, since the change transfer of change type "A" has been executed at the "200th day" of the date sequence, the "predicted date number" predicted by the time series model is 100 days (300-200=100). In other words, the "predicted date number" of this embodiment is related to the target date sequence. Based on this, as shown in Table 2, the processor 130 can obtain the change transfer requirement quantity and the predicted date number of change type "A". It is assumed that the processor 130 also executes the processes of the above-mentioned embodiments for the change type "B" and the change type "C". The processor 130 can obtain the change transfer requirement quantity and the predicted date number of the change type "B" as shown in Table 2, and obtain the change transfer requirement quantity and the predicted date number of the change type "C". Table 2 Change Type Change transfer quantity required Forecast date number A 1200 100 B 2000 250 C 1000 50

After the processor 130 obtains the forecast date number, the processor 130 can use the forecast date number and the preset advance date number to obtain the number of days to be transferred in advance, and transmit the change transfer request to the external bank electronic device 200 through the transceiver 120 when the number of days to be transferred in advance is required. Further, the change transfer request may include the required amount of change transfer. For example, the preset advance date number may be 7 days. That is, in this embodiment, the number of days to be transferred in advance for the change type "A" is 93 days (the forecast date number "100 days" - the preset advance date number "7"). In one embodiment, the processor 130 may display a change transfer planning reminder message through the input/output device 140 when the number of dates to be transferred in advance is required, so as to notify the bank's management staff to plan the change transfer in advance.

In other embodiments, according to the predicted date numbers shown in Table 2, the processor 130 can obtain the priority order of change transfer in the order of change type "C", change type "A" and change type "B". In other words, since the predicted date number of change type "C" is the smallest, the processor 130 can know that among change type "A", change type "B" and change type "C", change type "C" will be prioritized for change transfer.

In other embodiments, the processor 130 may determine which type of change will be prioritized for change transfer based on the change transfer costs of performing change transfers from different external banks. The change transfer costs may be, for example, the distance between the bank where the server 100 is located and the different external banks, or the change shipping costs required to perform change transfers from different external banks.

In summary, the server disclosed in the present invention for managing change transfer based on a time series model can use the time series model to predict the amount of change, and notify the bank's management staff when the number of transfer dates is required in advance to plan the change transfer in advance. In this way, it can avoid the situation of temporarily transferring change from an external bank when the amount of change is insufficient, thereby improving the customer satisfaction of the bank.

100: Server for managing change transfer based on time series model 110: Storage medium 120: Transceiver 130: Processor 140: Input and output device 200: External bank electronic device 21, 22: Curve 22-1, 22-2, 22-3, 22-n: Points

FIG1 is a schematic diagram of a server for managing change inflow based on a time series model according to an embodiment of the present disclosure. FIG2 is a schematic diagram of curves representing multiple predicted change amounts for the day according to an embodiment of the present disclosure.

100: A server that manages small change inbound transactions based on a time series model

110: Storage media

120: Transceiver

130: Processor

140: Input and output devices

200: External banking electronic devices

Claims (4)

A server for managing change transfer based on a time series model, comprising: a storage medium; a transceiver, communicatively connected to an external bank electronic device; an input-output device; and a processor, coupled to the storage medium, the transceiver and the input-output device, wherein the processor establishes a time series model corresponding to the change type; the processor uses the time series model to obtain a target date sequence corresponding to the change type, wherein the target date sequence corresponds to a target predicted amount of change for the day, and the target predicted amount of change for the day is less than or equal to a change threshold; The processor obtains the number of dates to be transferred in advance using the predicted date number and the preset advance date number, and transmits a change transfer request to the external bank electronic device through the transceiver when the number of dates to be transferred in advance is reached, wherein the predicted date number is related to the target date sequence, wherein the variables of the time series model include date sequence, the amount of change before training, the amount of change exchanged in the past week, the amount of change deposited in the past week, the average amount in the past week, the average amount in the past month, and whether it is a transfer date sequence, wherein the time series model is an ARIMA model, a SARIMA model, or an LSTM (Long Short-Term Memory) model, wherein the processor displays a change transfer planning reminder message through the input/output device when the number of dates to be transferred in advance is reached. The server as described in claim 1, wherein the storage medium stores the date sequence corresponding to the change type, the amount of change on the training day corresponding to the change type, the amount of change on the day before the training corresponding to the change type, the amount of change exchanged in the past week corresponding to the change type, the amount of change deposited in the past week corresponding to the change type, the average amount in the past week corresponding to the change type, the average amount in the past month corresponding to the change type, and the date sequence corresponding to whether the change type is transferred in. A server as described in claim 2, wherein the date sequence includes a first date sequence, and the change amount on the training day includes a first change amount on the training day, wherein the first date sequence corresponds to the first change amount on the training day, and the first date sequence corresponds to a call-in date sequence, wherein the processor sets the first change amount on the training day as the change call-in request amount, and the change call-in request includes the change call-in request amount. A server as described in claim 1, wherein the processor obtains multiple predicted daily change amounts using the time series model, and the multiple predicted daily change amounts correspond to multiple date sequences, wherein the multiple predicted daily change amounts include the target predicted daily change amount, and the multiple date sequences include the target date sequence.

Images