JP7492407B2 - Business Support System - Google Patents

Description

The present invention relates to a business support system and a business support method that support managers who formulate business plans based on machine learning predictions.

Prediction systems that use machine learning neural networks, deep learning, multiple regression analysis, and other statistical prediction models to obtain predicted values are being put into practical use in a variety of fields. These prediction models are first given training data with complete values for input and output items in the learning stage, and the parameter values in the prediction model are adjusted to appropriate values so that the calculated values of the prediction model match as closely as possible. Then, in the prediction stage, the values of the input items are given to this prediction model and the predicted value is calculated.

Although the predictive model is identified so as to match the training data provided in the training phase as closely as possible, the predicted values are not always accurate. This is particularly noticeable when the quality or quantity of the training data is insufficient. Therefore, managers who use the predicted values as a reference to formulate business plans find it difficult to determine how much to trust the predicted values. Even if it is possible to roughly estimate the prediction error, or how much the predicted values will deviate, there are cases in which it is not possible to immediately formulate business plans, such as how to allocate resources accordingly.

As background technology in this technical field, there is, for example, technology such as that described in Patent Document 1. The sewage inflow prediction device in Patent Document 1 comprises, as components, a storage unit that stores predicted values and prediction errors, and a prediction step width determination unit. With this configuration, it is said that it is possible to arbitrarily change the length of time from the current time to the future time at which the sewage inflow is to be predicted.

The power forecasting method, device, and program in Patent Document 2 describe generating multiple sets of random number data for power data for each location, or multiple sets of random number data that reproduces power data for each time period at a location, or correlation coefficient values between two time periods, and compiling predicted values based on these as a frequency distribution. This allows power system operators to obtain the predicted range and accuracy of the total power generation output in the region.

Patent Document 3 also shows that in conventional prediction methods using regression models and neural networks, error bars and standard deviations are used to evaluate the error of future predicted values (Figure 13, etc.).

JP 2016-211877 A JP 2014-54048 A JP 2010-20442 A

However, the above-mentioned Patent Document 1 does not mention a function for showing the user the extent of the prediction error. Since the statistical values of the prediction error are not shown, the user does not know how likely it is that the predicted value will be off.

In addition, the above-mentioned Patent Document 2 is a frequency distribution of predicted values, not a frequency distribution of errors, so the user does not know how likely the predicted values are to be off.

Even when error bars showing maximum, minimum, or 3σ are used as in Patent Document 3, the distribution of prediction errors may be biased, and users cannot accurately grasp how far the predicted values are off.

Therefore, even if the techniques described in these patent documents are used, the manager who formulates the business plan cannot determine the degree to which the predicted values can be trusted. Furthermore, it is not possible to immediately formulate a business plan, such as how to allocate resources in the future, based on the predicted values.

The object of the present invention is to provide a business support system and method that supports the formulation of business plans based on machine learning predictions, in which the manager formulating the business plan can accurately determine the degree to which the predictions can be trusted, and in which the business plan can be formulated quickly.

In order to achieve the above object, the present invention provides a business support system for supporting the formulation of business plans, comprising: a memory unit for storing past predicted values and past actual values, or past prediction errors calculated from past predicted values and past actual values; a statistical analysis unit for statistically processing past prediction errors to obtain a prediction error statistical processing result; a prediction error reflected prediction value calculation unit for adding/subtracting the prediction error statistical processing result obtained by the statistical analysis unit to/from a future predicted value to obtain a prediction error reflected prediction value; a display unit for displaying the prediction error reflected prediction value obtained by the prediction error reflected prediction value calculation unit ; an automatic business plan calculation unit for superimposing a figure that is vertically movable on the display unit and indicates a user's estimated value and automatically calculates a future business plan in accordance with the position of the figure; and a screen for displaying the future business plan automatically calculated by the automatic business plan calculation unit, wherein data that will be affected by the future business plan automatically calculated by the automatic business plan calculation unit is displayed on the display unit .

According to the present invention, in a business support system and a business support method that support the formulation of business plans based on machine learning prediction values, a manager who formulates the business plan can accurately determine the degree to which the predicted values can be trusted, and a business support system and a business support method that enable rapid formulation of business plans can be realized.

Problems, configurations, and advantages other than those described above will become clear from the description of the embodiments below.

1 is a functional block diagram of a task support system according to a first embodiment of the present invention; 13 is an example of a display showing a histogram of prediction errors together with predicted values. 13 is an example of a histogram showing the frequency of forecast errors in one day. 13 is a display example showing a frequency distribution. FIG. 11 is a functional block diagram of a task support system according to a second embodiment of the present invention. This is an example of a display unit that displays a superimposed figure (square) that can be freely moved up and down. 13 is an example of a display unit showing an automatically calculated future business plan. 1 is an example of a display showing data that is affected by an automatically calculated future business plan.

Below, an embodiment of the present invention will be described with reference to the drawings. In each drawing, the same components are given the same reference numerals, and detailed descriptions of overlapping parts will be omitted.

The tasks that the business support system of the present invention can support include, specifically, cargo delivery tasks and plant and equipment operation tasks, but are not limited to these as long as they involve determining future business plans based on predicted values.

The business support system and business support method according to the first embodiment of the present invention will be described with reference to Figures 1 to 4.

Figure 1 is a diagram showing the overall configuration of the business support system of this embodiment. First, the past prediction errors 12 stored in the memory unit 10 are provided to the statistical analysis unit 14. In addition, the learning time memory unit 32, which stores date and time data when learning was performed in the past using a method such as machine learning, provides the learning time 34 of the prediction model to the specific period setting unit 28.

The specific period setting unit 28 displays the learning time 34 of the prediction model, and also displays a screen on which the administrator can refer to and input the specific period setting input value 30. The input specific period setting input value 30 is provided to the statistical analysis unit 14.

The statistical analysis unit 14 extracts data for a period corresponding to the specific period setting input value 30 from the past prediction errors 12 and performs statistical analysis. The resulting prediction error statistical processing result 16 is provided to the prediction error reflected prediction value calculation unit 18. The prediction error reflected prediction value calculation unit 18 is also provided with the future prediction value 20 predicted by the prediction unit 26. The prediction unit 26 outputs a future prediction value from a reference time.

The forecast error reflected forecast value calculation unit 18 adds/subtracts the forecast error statistical processing result 16 to/from the future forecast value 20 to obtain the forecast error reflected forecast value 22, which is provided to the display unit 24. The display unit 24 displays the forecast error reflected forecast value 22 on the screen. For example, a graph is displayed in which one axis is time (number of days) and the other axis is the future forecast error reflected forecast value 22. This allows the manager to know the forecast value that takes into account the statistical value of past forecast errors, enabling more accurate formulation of business plans.

The display unit 24 is also provided with a future forecast value 20, and also displays the future forecast value 20 before it is added or subtracted by the forecast error reflected forecast value calculation unit 18. This allows the manager to know the extent of the forecast error, and to accurately determine the degree to which the forecast value can be trusted.

In the above example, the past prediction error 12 was provided from the memory unit 10 to the statistical analysis unit 14 and used as is, but the memory unit 10 may store past predicted values and past actual values. In that case, the statistical analysis unit 14 may calculate the past prediction error 12 based on the past predicted values and past actual values as preprocessing, and perform statistical analysis on the calculated value.

Past actual values 45 stored in the actual value storage unit 43 are also provided to the display unit 24. Past predicted values 36 stored in the predicted value storage unit 42 are also provided to the display unit 24. Past actual values 45 and past predicted values 36 are displayed on the display unit 24. This allows the user to know how accurately future predicted values 20 have been correct in the past.

There may be multiple predicted values 36 for the same past point in time, depending on how far into the future the prediction is made. For example, if there is a prediction unit 26 that predicts future predicted values 20 for three time periods of 10 minutes, 20 minutes, and 30 minutes, then at 12:00 yesterday, three future predicted values 20 were obtained: 12:10, 12:20, and 12:30. Ten minutes later, at 12:10, three future predicted values 20 were obtained: 12:20, 12:30, and 12:40.

Displaying all of these future forecast values 20 on the display unit 24 would make the screen cluttered, so it is also a good idea to limit the information to only a portion of the information. This makes it easier to understand the information and reduces the risk of misunderstandings or misreadings. One way to limit the information is to keep the time length constant, as described below, but this is not the only way to narrow it down.

The target data time length setting unit 48 sets a fixed time length 50. The time length 50 set by the target data time length setting unit 48 is given to the display unit 24. For example, if the time length 50 is set to 10 minutes, the display unit 24 will display a future predicted value 20 predicted for 10 minutes from now at 12:20 in the location of 12:30 yesterday, but will not display a future predicted value 20 predicted for 20 minutes from now at 12:10 or a future predicted value 20 predicted for 30 minutes from now at 12:00 on the screen.

The prediction error statistical processing results 40 at past times stored in the prediction error statistical processing result storage unit 38 are provided to the past prediction error reflected predicted value calculation unit 44. The past prediction error reflected predicted value calculation unit 44 is also provided with the predicted value 36 at the past time from the predicted value storage unit 42. The past prediction error reflected predicted value calculation unit 44 calculates the past prediction error reflected predicted value 46 by adding or subtracting the prediction error statistical processing results 40 at the past time to the predicted value 36 at the past time, and provides this to the display unit 24. The display unit 24 displays this past prediction error reflected predicted value 46 on the screen. This makes it possible to know how the past prediction error reflected predicted value 46 has changed at past times.

Here, the past prediction error statistical processing result 40 given to the past prediction error reflected prediction value calculation unit 44 may be the result of statistical processing of the past actual value 45 stored in the actual value storage unit 43 and the past predicted value 36 stored in the predicted value storage unit 42.

Figure 2 is an example of the display screen of the display unit 24. The downward black triangle at the top of the figure indicates the horizontal axis (number of days) = 0 days, which indicates the current time point. Future forecast values 20 are displayed as dotted lines from 0 days on the horizontal axis (number of days) to 4 days. Forecast error-reflecting forecast values 22 are displayed as a histogram rotated 90 degrees left at the 1st, 2nd, 3rd, and 4th days on the horizontal axis (number of days). Meanwhile, in the area of the horizontal axis (number of days) from -4 days to 0 days, where the horizontal axis (number of days) has negative values, the trend from the past to the present is shown as a line, with the solid line indicating the actual value at the past point in time 45 and the dotted line indicating the forecast value at the past point in time 36.

In FIG. 2, many of the histogram points displayed on the horizontal axis (number of days) for 1, 2, 3, and 4 days are below the dotted line indicating the future predicted value 20. This means that the actual values obtained in the past have been lower than the future predicted value 20 calculated by the prediction model (prediction unit 26).

The histogram shown in Figure 2 is obtained using the following steps:

(1) First, the data to be subjected to statistical processing and drawn as a histogram is determined. To do this, a specific period setting input value 30 is input in the specific period setting unit 28. Here, the period of the data to be analyzed is set. The specific period setting unit 28 displays the learning period 34 of the prediction model. For example, the learning period 34 of the previous prediction model is displayed as 3 months ago, the learning period 34 of the prediction model before that as 1 year and 3 months ago, and the learning period 34 of the prediction model before that as 2 years and 6 months ago.

The administrator refers to this information and inputs the period of data to be the subject of the histogram in the specific period setting unit 28. For example, by setting the period from two months ago to one month ago, the one-month period after the previous learning time 34 of the prediction model can be set as the subject of analysis.

(2) The statistical analysis unit 14 performs statistical analysis on the past prediction errors 12 for the period of the specific period setting input value 30 set in (1). For example, the histogram displayed at "1 day" on the horizontal axis (number of days) in Figure 2 uses past prediction errors 12 predicted for a time length of just one day into the future from the present. An example of frequency distribution data for past prediction errors 12 is shown in Table 1.

Figure 3 shows the results of displaying the data in Table 1 as a histogram. Figure 3 shows that the minimum value of the data range for the forecast error of the number of cargoes is -50 [units/day] and the maximum value is -10 [units/day]. Since the forecast error is about -30 [units/day] on average and the frequency around this is high, it can be inferred that the future forecast value 20 is likely to deviate from the future forecast value 20 by about -30 [units/day] for just one day. Since the forecast value indicated by the dotted line at the 1st day on the horizontal axis (number of days) in Figure 2 is 235 [units/day], it can be inferred that the actual value is likely to be about 235-30=205 [units/day]. In this way, with the forecast value and Figure 3, it is possible to understand what the future forecast value 20 is likely to be.

To make it easier to understand at a glance, rather than as separate graphs, the histogram in Figure 3 has been rotated 90 degrees left and the value of the data interval for the prediction error has been placed at a position where 20 future prediction values have been added, resulting in the horizontal histogram in Figure 2.

The part of the histogram in Figure 3 with the highest frequency, -25 [pieces/day] to -35 [pieces/day], is displayed in Figure 2 at the vertical axis (number of cargoes) position of 210 (= 235-25) [pieces/day] to 200 (= 235-35) [pieces/day]. Similarly, the part with the lowest frequency, -10 [pieces/day] to -15 [pieces/day], is displayed in Figure 2 at the vertical axis (number of cargoes) position of 225 (= 235-10) [pieces/day] to 220 (= 235-15) [pieces/day].

By displaying a histogram rotated 90 degrees left like this, you can see at a glance that where the horizontal axis (number of days) is 1 day, the number of cargoes is likely to be around 205 (= 235 - 30) [pieces/day].

Histograms of forecast errors are not necessarily symmetrical as shown in Figure 3. For example, as shown at the point where the horizontal axis (number of days) is 4 days, the data section with high frequency may be biased to one side rather than in the center. If forecast errors are simply displayed using error bars, as in Patent Document 3, such bias cannot be recognized at all, and there is a risk of making an erroneous judgment. Therefore, by displaying the prediction errors in a histogram as shown in Figure 2, values with high frequency in the past become clear, allowing for a more appropriate judgment.

Prediction models trained with training data of insufficient quantity or quality generally have low accuracy, so the vertical width (prediction error) of the histogram shown in Figure 2 is large, and the deviation from the future predicted value of 20 is also large. As time passes, the amount of data gradually increases, and re-learning is repeated using training data that includes this data, resulting in increased accuracy, the width of the histogram shown horizontally in Figure 2 becoming smaller, and the deviation from the future predicted value of 20 becoming smaller.

By checking the shape and position of the histogram by switching the specific period setting input value 30 between after the previous learning, after the learning before last, and after the learning before that, the administrator can check how much accuracy has improved as learning data is accumulated and learning progresses.

In Figure 2, in the period from -4 to 0 days, where the horizontal axis (number of days) is a negative value, the solid line indicates actual values 45 at past times, and the dotted line indicates predicted values 36 at past times. The predicted values 36 at past times are values predicted for a time length of just one day at each time point. This figure shows that the actual values 45 at past times from -4 to 0 days were smaller than the predicted values 36 at past times.

In addition to the solid and dotted broken lines, for the period from -4 to 0 days when the horizontal axis (number of days) in Figure 2 is a negative value, the past forecast error reflected forecast value 46 may be superimposed and displayed as a horizontal histogram, just as for the period from 1 to 4 days when the horizontal axis (number of days) is a negative value. This makes it possible to see how the past forecast error reflected forecast value 46 at a past point in time has changed compared to the forecast value 36 at a past point in time.

As for the statistical analysis in the statistical analysis unit 14, in addition to the histograms mentioned above, methods of calculating maximum, minimum, average, mode, median, and standard deviation may be used if the past prediction errors 12 are the subject of the analysis. In either case, the results are added to or subtracted from the future prediction value 20 and displayed in the diagram in FIG. 2. In the case of maximum and minimum values, they are displayed as error bars. In the case of average, mode, and median, they are displayed as a single point or a single short line. When standard deviation is used, the value obtained by subtracting a constant multiple of the standard deviation from the average value and the value obtained by adding a constant multiple of the standard deviation to the average value are displayed as error bars. These histograms, error bars, single points, and single short lines may be displayed simultaneously, overlapping each other, rather than separately.

The frequency distribution displayed on the display unit 24 may be displayed, for example, as a histogram, color, color shading, width, length, line graph, curve, envelope, etc., as shown in FIG. 4.

Note that the prediction unit 26 in FIG. 1 includes at least one of a neural network, deep learning, and multiple regression calculation, but is not limited to these as long as it is a calculation method that uses machine learning using training data.

As described above, the business support system of this embodiment includes a memory unit (memory unit 10, forecast value memory unit 42, actual value memory unit 43) that stores past forecast values (forecast values at past times 36) and past actual values (actual values at past times 45), or past forecast errors 12 calculated from past forecast values and past actual values, a statistical analysis unit 14 that performs statistical processing on the past forecast errors 12 to obtain a forecast error statistical processing result 16, a forecast error reflected forecast value calculation unit 18 that adds/subtracts the forecast error statistical processing result 16 obtained by the statistical analysis unit 14 to/from a future forecast value 20 to obtain a forecast error reflected forecast value 22, and a display unit 24 that displays the forecast error reflected forecast value 22 obtained by the forecast error reflected forecast value calculation unit 18.

The business support method of this embodiment also includes the steps of (a) calculating past prediction errors 12 from past predicted values (predicted values at past times 36) and past actual values (actual values at past times 45), (b) statistically processing the past prediction errors 12 calculated in step (a) to obtain a prediction error statistical processing result 16, (c) adding or subtracting the prediction error statistical processing result 16 obtained in step (b) to a future predicted value 20 to obtain a prediction error-reflected prediction value 22, and (d) displaying the prediction error-reflected prediction value 22 obtained in step (c) on the display unit 24.

This allows managers who create business plans to easily estimate what future forecast values will be after understanding the prediction error of the forecasting section based on past performance, allowing them to use this information to create more appropriate business plans.

In addition, managers can interactively calculate future business plans by moving shapes to set their own estimated values, allowing them to quickly formulate business plans.

The business support system and business support method according to the second embodiment of the present invention will be described with reference to Figures 5 to 8.

Figure 5 is a diagram showing the overall configuration of the business support system of this embodiment. The business support system of this embodiment shown in Figure 5 adds graphic position information 52, automatic business plan calculation unit 54, business plan 56, and affected data 58 to the components shown in Example 1 (Figure 1). The other configurations are the same as those of Example 1 (Figure 1). Of these, graphic position information 52 will be explained using Figure 6.

In FIG. 6, squares are displayed on the horizontal axis (number of days) at the 1st, 2nd, 3rd, and 4th days. This figure shows the number of cargoes estimated by the manager, and can be moved up and down on the display unit 24 for each day by operating the mouse, keyboard, touch panel, etc. The shape of the figure is not limited to a square, and may be a different shape. The value on the vertical axis (number of cargoes) of the moved figure becomes the figure position information 52.

As shown in FIG. 5, the graphic position information 52 is provided from the display unit 24 to the automatic business plan calculation unit 54. The automatic business plan calculation unit 54 calculates a business plan 56 based on the graphic position information 52 and provides it to the display unit 24. In addition, data 58 that is affected by the business plan 56 is also provided to the display unit 24 and displayed.

An example of the business plan 56 displayed on the display unit 24 is shown in Figure 7. In this example, the business plan 56 to be drawn up is the number of trucks to be operated. Figure 7 shows how many trucks are required to deliver the amount of cargo shown in Figure 6. On the horizontal axis (number of days), from -4 days to 0 days, past actual values for the number of trucks in operation, including today, are displayed, and on the horizontal axis (number of days), from 1st to 4th days, corresponds to the future business plan 56.

There are various methods for calculating the business plan 56, but one simple method is to divide the cargo quantity set by the graphic position information 52 by the transport capacity per truck and round up the value to an integer. It is desirable for Figures 6 and 7 to be displayed simultaneously on the same screen, and it is desirable for the display of the business plan 56 in Figure 7 to change in real time in response to the operation of moving the graphic (square) in Figure 6 up and down.

The manager who formulates the business plan can estimate the number of cargoes while checking the future predicted value 20 of the number of cargoes and the prediction error reflected prediction value 22, which is obtained by adding or subtracting the frequency distribution of past prediction errors to that predicted value, in Figure 6, and can easily understand interactively how many trucks will be in operation if the estimated number changes.

If it is desirable to have fewer trucks in operation, then if the number of trucks in operation can be reduced by one by slightly reducing the guess number, then it is easy to formulate a plan under the reduced guess number. Conversely, if the number of trucks in operation can be increased by one by slightly increasing the guess number, then it is possible to formulate a plan under the increased guess number, taking into account the risks.

Figure 8 shows an example of automatically calculated data 58 affected by future business plans. As data 58 affected by future business plans, the trend in the number of cargo in stock in the warehouse is displayed in a line graph. As in Figure 7, the horizontal axis (number of days) shows past actual values of cargo in stock in the warehouse, including today, from -4 to 0 days. As in Figure 7, the horizontal axis (number of days) shows data 58 affected by future business plans 56 from 1 to 4 days. There are various methods for calculating affected data 58, but a simple method is to subtract the value obtained by multiplying the number of trucks in operation in Figure 7 by the transport capacity per truck from the number of cargo in Figure 6, and then add the cargo in stock in the warehouse up to the previous time.

It is desirable that FIG. 8 is also displayed simultaneously with FIG. 6 on the same screen, and that the display of data 58 affected by FIG. 8 changes in real time in response to the operation of moving the figure (square) in FIG. 6 up and down.

When there is not one item but multiple items of affected data 58, it is advisable to simultaneously display trend graphs of the affected data 58 for each item on the same screen. In the example of Figure 8, the cargo stock in the warehouse is displayed, but the manager formulating the business plan can move the figure (square) in Figure 6 up and down to see the change in the number of cargo stock in the warehouse, and can formulate a plan for the number of trucks to be operated so that the number of cargo stock in the warehouse is at an appropriate level, while keeping in mind the upper limit of the warehouse capacity.

As described above, according to this embodiment, the data affected by the automatically calculated business plan is also displayed, so in addition to the effect of embodiment 1, the estimated value can be adjusted while checking the displayed business plan and the affected data.

The present invention is not limited to the above-described embodiments, but includes various modified examples. For example, the above-described embodiments have been described in detail to aid in understanding the present invention, and are not necessarily limited to those having all of the configurations described. It is also possible to replace part of the configuration of one embodiment with the configuration of another embodiment, and it is also possible to add the configuration of another embodiment to the configuration of one embodiment. It is also possible to add, delete, or replace part of the configuration of each embodiment with other configurations.

SYMBOLS WITH SYMBOLS 10... MEMORY UNIT 12... PAST FORECAST ERROR 14... STATISTICAL ANALYSIS UNIT 16... FORECAST ERROR STATISTICAL PROCESSING RESULT 18... FORECAST ERROR REFLECTING FORECAST VALUE CALCULATION UNIT 20... FUTURE FORECAST VALUE 22... FORECAST ERROR REFLECTING FORECAST VALUE 24... DISPLAY UNIT 26... PREDICTION UNIT 28... SPECIFIC PERIOD SETTING UNIT 30... SPECIFIC PERIOD SETTING INPUT VALUE 32... LEARNING TIME MEMORY UNIT 34... LEARNING TIME OF PREDICTION MODEL 36... FORECAST VALUE AT PAST POINT 38... FORECAST ERROR STATISTICAL PROCESSING RESULT MEMORY UNIT 40... PAST FORECAST ERROR STATISTICAL PROCESSING RESULT 42... FORECAST VALUE MEMORY UNIT 43... ACCURACY VALUE MEMORY UNIT 44... PAST FORECAST ERROR REFLECTING FORECAST VALUE CALCULATION UNIT 45... ACCURACY VALUE AT PAST POINT 46... PAST FORECAST ERROR REFLECTING FORECAST VALUE 48... SUBJECT DATA TIME LENGTH SETTING UNIT 50... TIME LENGTH 52... GRAPH POSITION INFORMATION 54... WORK PLAN AUTOMATIC CALCULATION UNIT 56... WORK PLAN 58... AFFECTED DATA

Claims (12)

In a business support system that supports the formulation of business plans,
a storage unit that stores past predicted values and past actual values, or past prediction errors calculated from past predicted values and past actual values;
a statistical analysis unit that performs statistical processing of past prediction errors to obtain a prediction error statistical processing result;
a prediction error reflected prediction value calculation unit that calculates a prediction error reflected prediction value by adding/subtracting the prediction error statistical processing result obtained by the statistical analysis unit to/from a future prediction value;
a display unit that displays the prediction error reflected prediction value calculated by the prediction error reflected prediction value calculation unit;
an automatic work plan calculation unit that displays a graphic that is movable up and down on the display unit and indicates a user's estimated value in a superimposed manner and automatically calculates a future work plan according to the position of the graphic;
a screen displaying a future business plan automatically calculated by the business plan automatic calculation unit;
Equipped with
a task support system, characterized in that data that will be affected by the future task plan automatically calculated by said task plan automatic calculation unit is displayed on said display unit ; 2. The business support system according to claim 1,
A business support system comprising a prediction unit that outputs a predicted value in the future from a reference time. 3. The business support system according to claim 2,
The business support system is characterized in that the prediction unit is equipped with at least one of a neural network, deep learning, and multiple regression calculation. 4. The business support system according to claim 1,
The business support system, wherein the display unit displays a graph having one axis representing time and another axis representing a future forecast error-reflecting predicted value. 5. The business support system according to claim 1,
A business support system comprising a specific period setting unit capable of setting a period of past prediction errors to be subjected to statistical processing by the statistical analysis unit. 6. The business support system according to claim 5,
A learning time storage unit is provided for storing a prediction model learning time, which is the time when the prediction model was learned;
A business support system, characterized in that the prediction model learning time output from the learning time storage unit is displayed on the specific period setting unit. 7. The business support system according to claim 1,
A business support system, wherein the result of the prediction error statistical processing is a diagram showing a frequency distribution. The business support system according to claim 7,
A business support system characterized in that the diagram showing the frequency distribution is displayed using at least one of a histogram, color, color shading, width, length, line graph, curve, and envelope. 3. The business support system according to claim 2,
A business support system, characterized in that the display unit displays future predicted values output from the prediction unit. 10. The business support system according to claim 1,
The storage unit includes a predicted value storage unit that stores a predicted value at a past time point;
A performance value storage unit that stores performance values at past times,
a display unit displaying the predicted value at a past time point output from the predicted value storage unit and the actual value at a past time point output from the actual value storage unit. The business support system according to claim 10,
The storage unit includes a prediction error statistical processing result storage unit that stores a prediction error statistical processing result at a past time point;
a past-time prediction error reflected predicted value calculation unit that calculates a past-time prediction error reflected predicted value by adding/subtracting the prediction error statistical processing result at the past time point stored in the prediction error statistical processing result storage unit and the predicted value at the past time point stored in the predicted value storage unit,
The business support system further comprises displaying on the display unit the past-time forecast error reflected predicted value calculated by the past-time forecast error reflected predicted value calculation unit. The business support system according to claim 11,
There are multiple forecast values with different time lengths until the future time of the forecast target,
A task support system comprising a target data time length setting unit that specifies the time length of target data used in the past forecast error reflected forecast value calculation unit.

Images