This disclosure relates generally to managing products at a retail sales facility and, in particular, to systems and methods for forecasting on-shelf availability of products on a sales floor of the retail sales facility.
A sales floor of a typical retail sales facility such as a large department store may have hundreds of shelves and thousands of products on the shelves displayed to the consumers. Periodically, products are taken off the shelves and purchased by the consumers. To restock the shelves after products are purchased by the consumers, overstock products stored in the stock room of the retail sales facility are picked from their bins and worked to the shelves on the sales floor.
Retail sales facilities determine how many units of a given product are on a shelf on a sales floor by way of manually auditing the products on the shelves. Specifically, workers at the retail sales facility periodically walk the aisles on the sales floor and use hand-held scanners to scan the products stocked on the shelf to take inventory of the products. Given the large number of shelves at a typical retail sales facility and the large number of products on the shelves, such manual auditing of the products on the shelves is very time consuming and less effective for the workers at the retail sales facility and increases the costs of operation for the retail sales facility.
Disclosed herein are embodiments of systems, devices, and methods pertaining to methods and systems for estimating whether a product is present or not present on a shelf on a sales floor of a retail sales facility at a given time interval. This description includes drawings, wherein:
Elements in the figures are illustrated for simplicity and clarity and have not necessarily been drawn to scale. For example, the dimensions and/or relative positioning of some of the elements in the figures may be exaggerated relative to other elements to help to improve understanding of various embodiments of the present invention. Also, common but well-understood elements that are useful or necessary in a commercially feasible embodiment are often not depicted in order to facilitate a less obstructed view of these various embodiments of the present invention. Certain actions and/or steps may be described or depicted in a particular order of occurrence while those skilled in the art will understand that such specificity with respect to sequence is not actually required. The terms and expressions used herein have the ordinary technical meaning as is accorded to such terms and expressions by persons skilled in the technical field as set forth above except where different specific meanings have otherwise been set forth herein.
The following description is not to be taken in a limiting sense, but is made merely for the purpose of describing the general principles of exemplary embodiments. Reference throughout this specification to “one embodiment,” “an embodiment,” or similar language means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the present invention. Thus, appearances of the phrases “in one embodiment,” “in an embodiment,” and similar language throughout this specification may, but do not necessarily, all refer to the same embodiment.
Generally, this application describes systems and methods of forecasting on-shelf availability of products at a retail sales facility, and more specifically, forecasting whether a product is present or not present at a selected time on the shelf on the sales floor of a retail sales facility by processing one or more on-shelf prediction factors associated with the product.
In one embodiment, a system of forecasting on-shelf availability of at least one product at a retail sales facility includes an electronic inventory management electronic device including a processor-based control unit configured to obtain electronic data including at least one on-shelf prediction factor associated with the at least one product. The at least one on-shelf prediction factor comprises at least one of: on-shelf probability state of the at least one product at the retail sales facility, conditional probability of sale of the at least one product at the retail sales facility during the selected time interval, root mean square error for cumulative sales of the at least one product at the retail sales facility. The control unit is further configured to estimate, based on that at least one on-shelf prediction factor, whether the at least one product is present or not present on a shelf on a sales floor of the retail sales facility at a selected interval of time on a given day, and to output, based on the estimation, a signal including electronic data indicating whether the at least one product is present or not present on the shelf on the sales floor of the retail sales facility at the selected interval of time on the given day.
In another embodiment, a method of forecasting on-shelf inventory of products at a retail sales facility includes: obtaining, by an electronic inventory management device, electronic data including at least one on-shelf prediction factor associated with the at least one product, the at least one on-shelf prediction factor comprising at least one of: an on-shelf probability state of the at least one product at the retail sales facility, a conditional probability of sale of the at least one product at the retail sales facility during the selected time interval, and a root mean square error for cumulative sales of the at least one product at the retail sales facility; estimating, by a processor-based control unit of the electronic inventory management device and based on the at least one on-shelf prediction factor, whether the at least one product is present or not present on a shelf on a sales floor of the retail sales facility at a selected interval of time on a given day; and outputting, by the electronic inventory management device and based on the estimating step, a signal including electronic data indicating whether the at least one product is estimated to be present or not present on a shelf on a sales floor of the retail sales facility at a selected interval of time on a given day.
The system 100 depicted in
It will be appreciated that the electronic inventory management device 120 may be configured for wired or wireless communication with one or more electronic devices (e.g., database server, regional server, or the like) located at or remote to the retail sales facility 110 and configured for two-way communication with the electronic inventory management device 120. It will also be appreciated that the electronic inventory management device 120 may be itself located remote to the retail sales facility 110 and configured for communication with one or more stationary or portable electronic devices local to the retail sales facility 110.
With reference to
The electronic data representing the real-time inventory information stored in the inventory management database 140 may include historical data derived from transaction data (e.g., sales) and worker task data (e.g., delivery, binning, and/or picking) associated with the products 190, as well as data indicating total number of products 190 in inventory and maximum shelf space for the products 190 at the retail sales facility 110. For example, the real time inventory data may include a total number of products 190 available in the retail sales facility 110 at a given time or historically over a period of one or more days or one or more weeks. The historical sales information stored in the inventory management database 140 may include the total number of products 190 sold at the retail sales facility 110 historically over a period of one or more intervals during a day, one or more days, or one or more weeks. The on-shelf estimation data stored in the inventory management database 140 may include data generated based on physical audits of shelves 180 at the retail sales facility 110 containing the products 190 for which an estimation of whether or not the products 190 are present on the shelf 180 were made by the electronic inventory management device 120.
In some embodiments, the inventory management database 140 may store electronic data in the form of on-shelf prediction factors. As discussed in more detail below, the on-shelf prediction factors are factored in by the processor of the electronic inventory management device 120 in estimating whether a product 190 is present or not present on a shelf 180 on the sales floor 170 of the retail sales facility 110 at a given time. Such on-shelf prediction factors will be discussed in more detail below and include, but are not limited to: on-shelf probability state of a product 190 at the retail sales facility 110; conditional probability of sale of the product 190 at the retail sales facility 110 during a selected time interval; root mean square error for cumulative sales of the product 190 at the retail sales facility 110; probability of sale of the product 190 at the retail sales facility 110 based on at least one interval of time equal to the selected interval of time but on at least one day prior to the given day; a consumer demand for the product 190 at the retail sales facility 110 during a predetermined time interval; average sales of the product 190 at the retail sales facility 110 during the predetermined time interval; average sales of the product 190 at the retail sales facility 110 based on a day of the week; a percentage of sales attributed to sales of the product 190 at the retail sales facility 110 within a product category associated with the product 190; format of the retail sales facility; travel time for replenishment of the product 190 at the retail sales facility 110; perpetual inventory of the product 190 at the retail sales facility 110; total sales of the product 190 at the retail sales facility 110 during at least one time interval preceding the selected interval of time on the given day; time elapsed since last sale of the product 190 at the retail sales facility 110; available space for the product 190 on the shelf 180 on the sales floor 170 of the retail sales facility 110; a type of the product 190; mod effective date; and at least one demographic variable associated with the retail sales facility 110.
The on-shelf prediction factors and other electronic data that may be stored in the inventory management database 140 in association with the products 190 at the retail sales facility 110 may be received by the electronic inventory management device 120, for example, as a result of a worker (e.g., stock room associate) scanning the products 190 using the scanning device 130, for example, during binning of the product 190 or when placing the product 190 onto a shelf 180. In some embodiments, at least some of the electronic data representing one or more of the on-shelf prediction factors may be transmitted to the electronic inventory management device 120 from the point-of-sale device 185 (e.g., sale register) local to the retail sales facility 110 or from one or more databases remote to the retail sales facility 110.
It will be appreciated that the inventory management database 140 does not have to be incorporated into the electronic inventory management device 120 as shown in
In some embodiments, the scanning device 130 of
After a product 190 is scanned via the scanning device 130 as described above, the electronic inventory management device 120 may receive electronic data associated with the product 190 (e.g., data uniquely identifying the product 190) from the scanning device 130 by way of a two-way communication channel 125, which may be a wired or wireless (e.g., Wi-Fi) connection. For example, when a worker places a product 190 onto a shelf 180 on the sales floor 170 of the retail sales facility 110, the worker may use the scanning device 130 to scan the unique identifier of the product 190, in response to which the data uniquely identifying the product 190 is obtained by the scanning device 130. In addition, as the worker places the product 190 into the shelf 180 on the sales floor 170, data identifying the task performed by the worker with respect to the product 190 (i.e., restocking) may be entered into the system 100 via the scanning device 130.
An exemplary electronic inventory management device 120 depicted in
This control unit 210 can be configured (for example, by using corresponding programming stored in the memory 220 as will be well understood by those skilled in the art) to carry out one or more of the steps, actions, and/or functions described herein. In some embodiments, the memory 220 may be integral to the processor-based control unit 210 or can be physically discrete (in whole or in part) from the control unit (i.e., control unit) 210 and is configured non-transitorily store the computer instructions that, when executed by the control unit 210, cause the control unit 210 to behave as described herein. (As used herein, this reference to “non-transitorily” will be understood to refer to a non-ephemeral state for the stored contents (and hence excludes when the stored contents merely constitute signals or waves) rather than volatility of the storage media itself and hence includes both non-volatile memory (such as read-only memory (ROM)) as well as volatile memory (such as an erasable programmable read-only memory (EPROM))).
Accordingly, the memory 220 and/or the control unit 210 may be referred to as a non-transitory medium or non-transitory computer readable medium. The control unit 210 of the electronic inventory management device 120 is also electrically coupled via a connection 235 to an input/output 240 that can receive signals from and send (via a wired or wireless connection) signals (e.g., commands, inventory database information) to devices (e.g., scanning device 130) local to the retail sales facility 110, or one or more devices remote to the retail sales facility 110.
Optionally, instead of receiving information regarding the products 190 in the bins 150 from a separate scanner such as the scanning device 130, the control unit 210 may incorporate or be electrically coupled to a sensor such as a reader configured to detect and/or read information on the identifying indicia (e.g., a label) located on the products 190 and/or on the bin 150 when the electronic inventory management device 120 is placed in direct proximity to the product 190 and/or the bin 150. Such an optional reader may be a radio frequency identification (RFID) reader, an optical reader, a barcode reader, or the like.
In the embodiment shown in
With reference to
The exemplary method 300 shown in
In the embodiment illustrated in
Some other exemplary on-shelf prediction factors that may be processed by the control unit 210 of the electronic inventory management device 120 to determine whether the product 190 is present or not on the shelf 180 on the sales floor 170 of the retail sales facility 110 include but are not limited to on-shelf probability state of a product 190 at the retail sales facility 110; conditional probability of sale of the product 190 at the retail sales facility 110 during a selected time interval; root mean square error for cumulative sales of the product 190 at the retail sales facility 110; probability of sale of the product 190 at the retail sales facility 110 based on at least one interval of time equal to the selected interval of time but on at least one day prior to the given day; a consumer demand for the product 190 at the retail sales facility 110 during a predetermined time interval; average sales of the product 190 at the retail sales facility 110 during the predetermined time interval; average sales of the product 190 at the retail sales facility 110 based on a day of the week; a percentage of sales attributed to sales of the product 190 at the retail sales facility 110 within a product category associated with the product 190; format of the retail sales facility; travel time for replenishment of the product 190 at the retail sales facility 110; perpetual inventory of the product 190 at the retail sales facility 110; total sales of the product 190 at the retail sales facility 110 during at least one time interval preceding the selected interval of time on the given day; time elapsed since last sale of the product 190 at the retail sales facility 110; available space for the product 190 on the shelf 180 on the sales floor 170 of the retail sales facility 110; a type of the product 190; mod effective date; and at least one demographic variable associated with the retail sales facility 110. Some of the above-listed on-shelf prediction factors are discussed in more detail below.
In the exemplary method 300 illustrated in
The “on-shelf probability state” on-shelf prediction factor for a product 190 reflects a probability of a given number of products 190 being on the shelf 180 at a given time, with the number of the products 190 on the shelf 180 being less than the maximum shelf space for the product 190 at the retail sales facility 110. In some embodiments, the control unit 210 of the electronic inventory management device 120 is programmed to assume that no product 190 can have more units on the shelf 180 than the allocated maximum capacity of the product 190 on the shelf 180 at any time of the day. The ‘state’ of the product 190 on the shelf 180 based on such an assumption can vary between 0 units to full shelf (i.e., maximum shelf capacity). The on-shelf probability state on shelf prediction factor (X1) at a given time t may be defined as shown below:
where Max k=shelf capacity and S is sale in time interval t
In some embodiments, the control unit 210 of the electronic inventory management device 120 is programmed to assign a probability to each of the on-shelf states of the product 190 based on sales of the products 190 at the retail sales facility 110. For example if the maximum shelf capacity of a product 190 is three units (as shown in
In some embodiments, the control unit 210 of the electronic inventory management device 120 is programmed to track sales of a product 190 at the retail sales facility 110 at every 15 minute interval, and to recalculate the change in the on-shelf probability state relative to the starting on-shelf probability at every 15 minute interval. If electronic data indicating a sale of one of the three products 190 is received by the electronic inventory management device 120 at 12:15 am following a 12:00 am start of the 15 minute interval, the control unit 210 of the electronic inventory management device 120 may be programmed to recalculate the on-shelf state probability from its initial value of 0.25 to a modified value of 0.33 (assuming no re stocking has taken place in the last interval), since only 3 possible states of the product 190 on the shelf 180 remain, since there would be either 2 or 1 or 0 products 190 on the shelf 180 at any given time following the sale of one of the three products 190 initially present on the shelf 180 assuming no restocking of the product 190 happened in the interval between 12:00 am to 12:15 am.
In some embodiments, the control unit 210 is programmed to interpret a value of the on-shelf probability state factor (X1t) being greater than 1 or less than 0 as an indication that a restocking of the product 190 on the shelf 180 has been carried out by a worker at the retail sales facility 110. The value of the probability state factor (X1t) is then reset to initial value calculated for the start of the day. The control unit 210 may also be programmed to interpret a higher values, closer to 1 but not greater than 1 or <0 of the calculated on-shelf state probability as an indication of a higher likelihood that the product 190 (which starts with a full shelf 180 at the start of the day) is not present on the shelf 180 for a product 190. Since it may be difficult to define a start of the day time for a retail sales facility 110 that operates 24 hours a day, in some embodiments, the control unit 210 of the electronic inventory management device 120 may be programmed to interpret the start of day time as the time when the product 190 is at maximum capacity on the shelf 180 on the sales floor 170 of the retail sales facility 110. It will be appreciated that since not all products 190 may be at full shelf capacity at the start of the day, and given that on-shelf state probabilities for products 190 having no consumer demand may peak at a certain value and not change, the control unit 210 may be programmed to evaluate one or more on-shelf prediction factors in addition to the on-shelf probability state factor in order to more accurately estimate whether the product 190 is present or not present on the shelf 180 at any given time throughout the day.
The “conditional probability of sale in an interval” on-shelf prediction factor for a product 190 refers to a probability of occurrence of a sale of the product 190 in a given interval of time given the known number of sales of the product 190 during the preceding identical interval. For example, based on this factor, the control unit 210 of the electronic inventory management device 120 may be programmed to interpret that a product 190 with a known forecasted demand of F for the given day so far is expected to have a sales volume approximately equal to based on historical demand during any 15 minute intervals during the given day. In some embodiments, the control unit 210 of the electronic inventory management device 120 may be programmed to define the conditional probability of sale factor (X3) as:
where n=1 when St=0 & F is daily sales volume forecast
For example, the control unit 210 may be programmed to divide an entire day (i.e., 24 hour interval) into 96 intervals of 15 minute each, with the day starting at 12:00 am. Then, since the products 190 at the retail sales facility 110 undergo a finite number of unit sales at the point-of-sale device 185 in a given day (and during a given 15 minute interval of the day), and since each sale of the product 190 is transmitted from the point-of-sale device 185 to the electronic inventory management device 120 and recorded in the inventory management database 140, the control unit 210 can calculate the total number of sales of the product 190 throughout a given day for estimating daily forecast F and during any of the 96 15-minute intervals of the day for estimating cumulative volume sales till the start of any given interval of time, for which ‘Conditional probability of Sale in an interval’ on shelf prediction factor is to be calculated. Then based on the known total number of sales of the product 190 during the preceding 15-minute interval and the daily forecast, forecast of the sales for the next 15-minute interval of the day is estimated.
As seen above, a forecast by the control unit 210 based on the conditional probability of sale in time interval on-shelf prediction factor provides an approximation of the number of sales of a product 190 that can be expected across an interval of interest across a given day. As such, based on an assumption that each interval of time throughout the day has an equal probability of getting a sale of the product 190, the control unit 210 can estimate the probability of sale of the product 190 at a given interval knowing how many unit sales of the product 190 occurred in the preceding identical interval of time and the daily forecast based on historical demand. In some embodiments, the control unit 210 is programmed to interpret a high value of the conditional probability factor as an indication of an increased likelihood that the product 190 is not present on the shelf 180.
It will be appreciated that generally, not all intervals of the day have an equal probability of a sale of the product 190 taking place, since the arrival of customers at a retail sales facility 110 is not uniform throughout the day. Accordingly, some intervals of time throughout the day are likely to have a higher probability of sale than other intervals based on customer arrival distribution. As such, the control unit 210 of the electronic inventory management device 120 may be programmed to evaluate one or more on-shelf prediction factors in addition to the conditional probability of sale in a time interval factor in order to more accurately estimate whether the product 190 is present or not present on the shelf 180 at any given time throughout the day.
The “root mean square error for cumulative sales” on-shelf prediction factor refers to a variation in cumulative sales over an average cumulative sale is computed for every 15 minute interval of a day over an interval of one or more consecutive days (e.g., 7 days, 14 days, 30 days, 60 days, etc.). In other words, the root mean square error for cumulative sales on-shelf prediction factor is premised on an assumption that the average volume of sales of the product 190 over a selected daily time interval (e.g., 11:00 am to 11:15 am or 7:30 pm to 7:45 pm) during a selected period of days/weeks (1 week, 2 weeks, 4 weeks, 6 weeks, 8 weeks, etc.) indirectly indicates the possibility of sale of that product 190 occurring during the same time interval of the day (i.e., 11:00 am to 11:15 am or 7:30 pm to 7:45 pm) on the day for which the forecast is being made. In some embodiments, the control unit 210 of the electronic inventory management device 120 may be programmed to define the root mean square error for cumulative sales (X11) on-shelf prediction factor as:
X11=√{square root over ((Σstavg−Σst2)}
In some embodiments, the control unit 210 is programmed to determine the value representing an average number of unit sales of the product 190 during a given historical period (e.g., 8 weeks) by retrieving historical data relating to sales of the product 190 from the inventory management database 140. This determination by the control unit 210 is generally based on an assumption that for an average day, the number of units of product 190 sold during any interval is close to the historical average of sales for that product 190 during that interval of the day. Generally, products 190 having a high sales volume stay closer to this assumption in each interval, while slow-moving products 190 are further away from this assumption due to sales variations. Nonetheless, all products 190 are subject to fluctuation in the daily number of sales throughout a week (e.g., Monday vs. Sunday or regular day vs. holiday). Accordingly, some days throughout the week are likely to have a higher probability of sale of the product 190 during a specific time interval than other days during that same time interval based on customer arrival distribution. As such, the control unit 210 of the electronic inventory management device 120 may be programmed to evaluate one or more on-shelf prediction factors in addition to the root mean square error for cumulative sales on-shelf prediction factor in order to more accurately estimate whether the product 190 is present or not present on the shelf 180 at a given time interval of the day.
The “probability of sale” on-shelf prediction factor refers to the probability of sale of a product 190 during any 15 minute interval over a selected period of days or weeks (e.g., 8 week or 10 week average). In some embodiments, when estimating whether a product 190 is present or not on the shelf 180 based on the probability of sale on-shelf prediction factor, the control unit 210 of the electronic inventory management device 120 is programmed to obtain a historical value of sales of the product 190 in any given 15 minute interval of the day during the preceding 8 weeks and to evaluate each 15-minute interval independently of other 15-minute intervals in a day based on 8 week history sales for a given 15 minutes interval. As such, if the control unit 210 of the electronic inventory management device 120 retrieves (e.g., from the inventory management database 140) historical data indicating, for example, that 0.75 is the probability of sale of the product 190 during a 15-minute time interval based on the volume sale recorded for that interval in the last 8 weeks preceding the day for which the forecast is being made, the control unit 210 is programmed to assume that the probability of making a sale of at least 1 unit of product 190 during the forecasted 15 minute interval on the given day should be close to 3 out of 4 instances based on 8 weeks history In an another example, if the value of probability of sale is calculated to be 1 based on preceding 8 weeks of history data for any given interval, in this scenario, the control unit 210 would forecast that the product 190 is highly likely to make a sale during the 15 minutes interval for which the forecast is being made on a given day.
As discussed above, it will be appreciated that fluctuations in the sales of the product 190 at the retail sales facility 110 may occur throughout the course of a day and throughout the course of the week, based on customer arrival distribution. As such, the control unit 210 of the electronic inventory management device 120 may be programmed to evaluate one or more on-shelf prediction factors in addition to the “probability of sale” on-shelf prediction factor in order to more accurately estimate whether the product 190 is present or not present on the shelf 180 at a given time interval of the day.
The “product interval demand” on-shelf prediction factor refers to a consumer demand for a product 190 during a given interval of time. For example, in order to estimate a demand for a given interval of time (e.g., 15 minutes), in some embodiments, the control unit 210 of the electronic inventory management device 120 may be programmed to calculate/retrieve (e.g., from the inventory management database 140) a daily demand forecast for the product 190 at a level of the retail sales facility 110 and split this daily-level demand into the constituent intervals. Then, to estimate the availability or unavailability of the product 190 on the shelf 180 at the retail sales facility 110 at every 15 minute interval, the control unit 210 retrieves the known demand for the product 190 during each of these intervals and compares the interval demand with the actual number of units of the product 190 sold during the same interval.
In some embodiments, the control unit 210 of the electronic inventory management device 120 is programmed to set the daily demand as D and evaluate D as being directly proportional to the total transactions (T) in a given day. Since transactions (sales of the product 190) at the retail sales facility 110 are not constant, but continuously vary by time interval based on customer shopping patterns, the control unit 210 may be programmed to interpret the transactions associated with the product 190 as a function of time as follows:
T=f(t)
D∝g(T)∝g(f(t))
Then, the control unit 210 may be programmed to calculate the demand for the product 190 in an interval between tn and t(n+Δn) using the area under a curve defined by the following Equation
It will be appreciated that the demand for the product 190 at each interval is a Markov process and for a short interval can be represented using Poisson distribution. It will also be appreciated that the shape of the curve defined in Equation 1 above is expected to remain the same for all types the products 190 at a given retail sales facility 110, but the amplitude of the curve would be expected to vary based on the value of demand for the product 190. In some embodiments, the control unit 210 may be programmed to introduce a correction factor K in order for Equation 1 to reflect a more accurate variation of demand for a given product 190. An exemplary derivation of the correction factor K is discussed below. The correction factor K may take into account a correction based on customer behavior and product behavior at the retail sales facility 110.
The “product interval sales velocity” on-shelf prediction factor reflects accounts for the possibility that, for a given time interval, a product 190 may sell faster or slower as compared to other products 190 in the category of the product 190, or as compared to aggregated sales of products 190 at the retail sales facility 110. In some embodiments, the control unit 210 may be programmed to process this factor to estimate the correction factor K and to categorize a product 190. Generally, the product interval sales velocity on-shelf prediction factor may account for the difference in sales velocity of a product 190 under consideration when compared to a sales velocity of all other products 190 at the retail sales facility 110, as summarized by the equations below:
The “product day sales index” on-shelf prediction factor accounts for the possibility that a product 190 may move faster or slower relative to its average sales in an interval value based on which day of the week it is. As discussed above, average sales for a product vary from day to day throughout the week, with sales being higher, for example on the weekend (i.e., Friday night, Saturday, and Sunday) as compared to the week days (i.e., Monday-Thursday). In some embodiments, the control unit 210 may be programmed to process this factor to estimate the correction factor K and to categorize products 190 based on sales variation introduced due to sales pattern differences on different days of the week, as summarized by the equation below.
The “day of the week” on-shelf prediction factor accounts for the possibility that sales volume of a given product 190 and consumer demand for the product 190 at the retail sales facility 110 may be different on different days of the week.
The “sales category contribution” on-shelf prediction factor accounts for the share represented by sales of the product 190 being forecast relative to sales of other products 190 within the product category of the product 190. Another, related on-shelf prediction factor that may be factored in by the control unit 210 of the electronic inventory management device 120 when forecasting whether a product 190 is or is not present on the shelf 180 may be “sales category contribution standard deviation.” The “sales category contribution” and the “sales category contribution standard deviation” factors may vary between different retail sales facilities 110 and between different regions.
The “store format” on-shelf prediction factor may account for differences in the format between different retail sales facilities 110 where a forecast of whether a product 190 is or is not on the shelf 180 is made. For instance, one such difference in format maybe a retail sales facility 110 that has varying hours of operation (e.g., 9 am to 9 pm, 10 am to 6 pm, etc.) on different days of the week versus a 24-hour retail sales facility 110.
Another on-shelf prediction factor that may be factored in by the control unit 210 of the electronic inventory management device 120 when forecasting whether a product 190 is or is not present on the shelf 180 may be the “replenishment distance/travel time” factor (Ø), which takes into account an approximate distance from the location where the product 190 is displayed on the sales floor 170 at the retail sales facility 110 to the location of the product in the stock room 160. The distance may be normalized by the size of the retail sales facility 110 as follows:
Another on-shelf prediction factor that may be factored in by the control unit 210 of the electronic inventory management device 120 when forecasting whether a product 190 is or is not present on the shelf 180 may be the “perpetual inventory” factor, which is based on a snap shot of available inventory of a product 190 at a given point of time or interval of time at the retail sales facility 110. For example, the perpetual inventory factor may be obtained by the control unit 210 of the electronic inventory management device by querying the inventory management database 140. In some instances, the perpetual inventory factor may be subject to inaccuracies due to loss of products 190 at the retail sales facility 110 due to damage, shrinkage (miscounting, non-delivery, theft, damages etc.).
Other on-shelf prediction factors that may be factored in by the control unit 210 of the electronic inventory management device 120 when forecasting whether a product 190 is or is not present on the shelf 180 may include: product sales standard deviation, product day sales standard deviation (c); time elapsed since last sale of the product 190; shelf space (ki) for the product 190; whether the product 190 is primary or linked; product type; mod effective date; and demographic variables (e.g., location of the retail sales facility 110).
In some embodiments, the control unit 210 of the electronic inventory management device 120 may be programmed to compute the above-discussed correction factor K as follows:
K=ρ×μ (4)
In some embodiments, the control unit 210 is programmed to facilitate ease of interpretation and/or standardization of results by interpreting a result that Ein>ki for a interval tn, to conclude that the shelf 180 where the product 190 is displayed needs at least 1 cycle of restocking in the time interval tn to tn+1. Then, for a time interval tn to tn+1, if the below statement is True
Sin≧Ein>ki (5)
then,the control unit 210 is programmed to conclude that restocking of the shelf 180 took place in the given interval for the product and probability of unavailability of product 190 is low. Similarly, if the below statement is True
Ein≧ki>Sin (6)
then the control unit 210 is programmed to conclude that the probability of unavailability of the product 190 on the shelf 180 is high.
It will be appreciated that a fast-moving product 190 may undergo multiple rounds of restocking in a given time interval, depending on the size of the time interval. For example, in a short time interval (e.g., 15 minutes), the product 190 may or may not undergo restocking, while in a 12 hour time interval, the product 190 is more likely than not to be restocked. As such, in some embodiments, the control unit 210 of the electronic inventory management device 120 may interpret the restocking of a product 190 on the shelf 180 as a binary value, and may factor in multiple restocking events, if a time larger interval (e.g., 4 hours, 6, hours, or 8 hours) is chosen.
Equation 7 below represents another possible scenario that may be processed by the control unit 210 in order to estimate the probability of whether the product 190 is unavailable on the shelf 180:
Ein≧Sin>ki (7)
Equations 5, 6, and 7 above indicate that the relationship between actual sales (Sin), estimated demand for the product (Ein) and shelf space (ki) significantly affects the estimation of whether a product 190 is present or not on the shelf 180 at the retail sales facility 110. The interdependence of these features may be represented using a normalized single feature called velocity ratio defined below
If δ<0, this situation is equivalent to equation 5
and if δ>1, this situation is equivalent to equation 6
and if 0<δ1, this situation is equivalent to equation 7
It will be appreciated that Equations 5, 6, 7, and 8 may require modification to account for multiple restocking possibilities during a large time interval. The lower limit is then:
Limit=R*ki
where R is the number of restocking required for the interval
In addition, lost sales can be computed when the control unit 210 of the electronic inventory management device 120 correctly estimates that a product 190 is not present on the shelf 180 on the sales floor 170 of the retail sales facility 110 at a given time. The lost sales may be computed as follows:
In the exemplary embodiment of
Generally, the system 400 includes two subsystems, an algorithm training sub-system 405 and a product on-shelf availability prediction system 410. While
The exemplary algorithm training sub-system 405 shown in
As shown in
In the embodiment of
In some embodiments, the processor of the algorithm feature calculation and training application 425 may be programmed to develop a joint probability distribution function for the on-shelf prediction factors in a training set. Since the data sets are normalized, the processor may be programmed to assume that the data sets represent multi-dimensional normal distribution. The normal probability values of each instance in the training data set can be defined as
=N(x1, μ1, σ1), N(x2, μ2, σ2)N(x3, μ3, σ3). . . N(xn, μn, σn)
where μ and σ are mean and standard deviation of the feature data in the training set.
In some embodiments, the processor of the algorithm feature calculation and training application 425 may be programmed to use a separate data set for testing the effectiveness of the algorithm. For example, processor of the algorithm feature calculation and training application 425 may be programmed to as being effective if it is able to separate the instances into “on shelf” (e.g., value 1) and “not on shelf” (e.g., value 0). The electronic data representing the test set used to test the algorithm may be the historical transaction data obtained from the retail sales facility 110. Since this test set includes both “on shelf” and “not on shelf” instances, to test the algorithm, the processor of the algorithm feature calculation and training application 425 may be programmed to compute the probability values of each of the data points using the joint probability distribution as follows:
test
=N(x1test, μ1, σ1), N(x2test, μ2, σ2)N(x3test, μ3, σ3). . . N(xntest, μn, σn)
The processor of the algorithm feature calculation and training application 425 may be programmed to compute a threshold value of the probability for the distribution function developed using the training set. If this threshold is represented a ω, then for a condition, where test<ω, the prediction by the processor of the algorithm feature calculation and training application 425 is 0 (i.e., product 190 not on the shelf 180), and for a condition, where test>ω, the prediction by the processor of the algorithm feature calculation and training application 425 is 1 (i.e., product 190 is on the shelf 180).
The exemplary product on-shelf availability prediction sub-system 410 includes a historical data database 440, which, similarly to the historical data database 415, may store historical data including but not limited to historical sales volume of a product 190 at the retail sales facility 110 by time interval, maximum shelf capacity for the product 190 at the retail sales facility 110, and total inventory of products 190 at the retail sales facility 110. The on-shelf availability prediction sub-system 410 of
As shown in
The electronic data representing the estimation by the processor of the prediction management device 455 as to whether the product 190 is present on the shelf 180 or not may be transmitted to the retail sales facility inventory device 460 (e.g., formatted as a visual or audible alert), which in turn may forward this electronic data to the scanning device 430. The scanning device 430 may in turn generate a visual and/or audible alert to a worker at the retail sales facility indicating whether the product 190 of interest is present on the shelf 180 at a given time interval or not, enabling the worker to take appropriate action based on the alert. Such an action by the working may be picking more units of the product 190 from the bin 150 in the stock room 160 and bringing the picked units of the product 190 to the sales floor 170 for restocking the shelf 180 with the product 190.
As shown in
The systems and methods described herein analyze one or more on-shelf prediction factors to estimate whether a product is present or not present on a shelf on a sales floor of a retail sales facility at a given time or during a given time interval. Such estimation of whether or not the product is or is not present on the shelf on the sales floor of the retail sales facility advantageously alleviates the need to have workers at the retail sales facility to manually audit the products on the shelves multiple times a day, enabling the workers to perform other tasks that may be more needed.
Those skilled in the art will recognize that a wide variety of other modifications, alterations, and combinations can also be made with respect to the above described embodiments without departing from the scope of the invention, and that such modifications, alterations, and combinations are to be viewed as being within the ambit of the inventive concept.
Number | Date | Country | Kind |
---|---|---|---|
7065/CHE/2015 | Dec 2015 | IN | national |
This application claims the benefit of U.S. Provisional Application No. 62/293,644, filed Feb. 10, 2016, and relates to Indian Application No. 7065/CHE/2015, filed Dec. 30, 2015, which applications are incorporated by reference herein in their entireties.
Number | Date | Country | |
---|---|---|---|
62293644 | Feb 2016 | US |