Server and method for facilitating providing recommendation to merchant for on-demand service

Abstract

Aspects concern a server for facilitating providing a recommendation to a merchant for an on-demand service, the server comprising: a memory configured to store instructions; and a processor for executing the stored instructions and configured to: for each merchant of a plurality of merchants, classify the merchant into at least one cluster of a plurality of clusters based on one or more criteria relating to the merchant; collect information about a plurality of features of two or more merchants belonging to the cluster that the merchant belongs to; predict normalised sales based on the collected information about the plurality of features of the two or more merchants, using a machine learning model; calculate a Shapley value of each feature of the plurality of features, based on the predicted normalised sales, using the machine learning model; and generate the recommendation based on the calculated Shapley value.

Description

SERVER AND METHOD FOR FACILITATING PROVIDING RECOMMENDATION TO MERCHANT FOR ON-DEMAND SERVICE TECHNICAL FIELD

[0001] Various embodiments relate to a server and a method for facilitating providing a recommendation to a merchant for an on-demand service.BACKGROUND

[0002] In a competitive landscape of an on-demand service industry (for example, a food delivery industry), merchants (also referred to as “item providers”) (for example, restaurants) may be inundated with vast amounts of data and insights related to their sales, operations, and menu offerings. An on-demand service platform provider may provide the merchants with various analytics tools designed to help them assess their performance and identify areas for improvement. However, the sheer volume of information may be overwhelming, especially for small to medium-sized enterprises (SMEs) that may lack time, resources, or expertise to interpret these insights effectively. Additionally, the merchants may often struggle to understand which specific actions they can take to quickly boost their revenue and sales. For example, the merchants may face several key challenges in leveraging data for business optimisation, as follows:• Information Overload: The merchants may receive insights from multiple sources but lack clear guidance on which actions will yield the greatest impact on their sales. • Limited Time: The operational demands of running a restaurant may leave little room for in-depth data analysis.• Lack of Expertise: The merchants may not possess the technical knowledge required to interpret complex analytics effectively.• Competitive Pressure: To remain competitive, the merchants may need to continuously optimise their offerings based on market trends and competitor performance, which may be difficult without targeted guidance.

[0003] Existing solutions may attempt to address these challenges but have limitations as follows:• One-Size-Fits-All Recommendations: Generic suggestions may not be tailored to the merchants’ specific business circumstances.• Standard Analytics Dashboards or Rule-Based Models: While dashboards may display various performance metrics, they may lack contextual prioritization. Additionally, rule-based systems may provide root cause identification for declining sales but may not offer dynamic and tailored recommendations.• Manual Consulting Services: While third-party consultants may offer personalised advice, their services may be costly and not be scalable for smaller merchants.

[0004] In addition, these existing approaches may present inefficiencies, as follows:• Time-Consuming Decision-Making: Merchants may need to manually decipher which insights may be most relevant, taking valuable time away from operations.• Limited Effectiveness: Generic recommendations may fail to address specific needs of individual merchants, leading to suboptimal outcomes.• Accessibility Barriers: Personalised consulting services may be unaffordable for smaller businesses.

[0005] Therefore, there is a need to provide a solution for facilitating providing a recommendation to a merchant for the on-demand service. Specifically, there is a need to provide the solution that:Prioritises recommendations based on their potential impact on the merchants’ sales and revenue.• Provides customised recommendations by benchmarking against peer and competitor performance.• Learns and adapts recommendations over time based on the merchants’ activity and feedback.• Simplifies complex data into clear and actionable insights that can be easily implemented.• Enhances market competitiveness by empowering the merchants with data-driven strategies without requiring significant time investment or technical expertise.SUMMARY

[0006] According to various embodiments, there is a server for facilitating providing a recommendation to a merchant for an on-demand service, the server comprising: a memory configured to store instructions; and a processor for executing the stored instructions and configured to: for each merchant of a plurality of merchants, classify the merchant into at least one cluster of a plurality of clusters based on one or more criteria relating to the merchant; collect information about a plurality of features of two or more merchants belonging to the cluster that the merchant belongs to; predict normalised sales based on the collected information about the plurality of features of the two or more merchants, using a machine learning model; calculate a Shapley value of each feature of the plurality of features, based on the predicted normalised sales, using the machine learning model; and generate the recommendation based on the calculated Shapley value.

[0007] In some embodiments, the processor is further configured to: determine if a number of the two or more merchants belonging to the cluster is less than a threshold; and if it is determined that the number of the two or more merchants belonging to the cluster is less than the threshold, remove at least one criterion from the one or more criteria for re-classifying the merchant.

[0008] In some embodiments, the processor is further configured to: collect the information about the plurality of features of the two or more merchants over a past predetermined period; and aggregate the collected information about the plurality of features of the two or more merchants over the past predetermined period.

[0009] In some embodiments, the processor is further configured to: normalise the collected information about the plurality of features of the two or more merchants.

[0010] In some embodiments, the processor is further configured to: apply a standard z-scaling to the collected information about the plurality of features of the two or more merchants, to standardise the collected information about the plurality of features of the two or more merchants.

[0011] In some embodiments, the processor is further configured to: train the machine learning model to predict the normalised sales based on the collected information about the plurality of features of the two or more merchants.

[0012] In some embodiments, the processor is further configured to: determine a contribution of each feature of the plurality of features to the predicted normalised sales based on the calculated Shapley value; and generate the recommendation based on the determined contribution of each feature.

[0013] In some embodiments, the processor is further configured to: extract at least one feature with a negative Shapley value from the plurality of features, as an adjustable feature.

[0014] In some embodiments, the processor is further configured to: estimate a potential sales gain by re-calculating a Shaplcy value using the adjustable feature and identifying a difference between the calculated Shapley value and the re-calculated Shapley value; and if there are two or more adjustable features, rank the two or more adjustable features based on the corresponding potential sales gain, and select at least one adjustable feature from the two or more adjustable features based on the ranking, to generate the recommendation.

[0015] In some embodiments, the processor is further configured to: generate the recommendation using a Large Language Model (LLM).

[0016] According to various embodiments, there is a method for facilitating providing a recommendation to a merchant for an on-demand service, the method comprising: for each merchant of a plurality of merchants, classifying the merchant into at least one cluster of a plurality of clusters based on one or more criteria relating to the merchant; collecting information about a plurality of features of two or more merchants belonging to the cluster that the merchant belongs to; predicting normalised sales based on the collected information about the plurality of features of the two or more merchants, using a machine learning model; calculating a Shapley value of each feature of the plurality of features, based on the predicted normalised sales, using the machine learning model; and generating the recommendation based on the calculated Shapley value.

[0017] In some embodiments, the method further comprises: determining if a number of the t 'o or more merchants belonging to the cluster is less than a threshold; and if it is determined that the number of the two or more merchants belonging to the cluster is less than the threshold, removing at least one criterion from the one or more criteria for re-classifying the merchant.

[0018] In some embodiments, the method further comprises: collecting the information about the plurality of features of the two or more merchants over a past predetermined period; andaggregating the collected information about the plurality of features of the two or more merchants over the past predetermined period.

[0019] In some embodiments, the method further comprises: normalising the collected information about the plurality of features of the two or more merchants.

[0020] In some embodiments, the method further comprises: applying a standard z-scaling to the collected information about the plurality of features of the two or more merchants, to standardise the collected information about the plurality of features of the two or more merchants.

[0021] In some embodiments, the method further comprises: training the machine learning model to predict the normalised sales based on the collected information about the plurality of features of the two or more merchants.

[0022] In some embodiments, the method further comprises: determining a contribution of each feature of the plurality of features to the predicted normalised sales based on the calculated Shapley value; and generating the recommendation based on the determined contribution of each feature.

[0023] In some embodiments, the method further comprises: extracting at least one feature with a negative Shapley value from the plurality of features, as an adjustable feature.

[0024] In some embodiments, the method further comprises: estimating a potential sales gain by re-calculating a Shapley value using the adjustable feature and identifying a difference between the calculated Shapley value and the re-calculated Shapley value; and if there are two or more adjustable features, ranking the two or more adjustable features based on the corresponding potential sales gain, and selecting at least one adjustable feature from the two or more adjustable features based on the ranking, to generate the recommendation.

[0025] In some embodiments, the method further comprises: generating the recommendation using a Large Language Model (LLM).

[0026] According to various embodiments, a data processing apparatus configured to perform the method of any one of the above embodiments is provided.

[0027] According to various embodiments, a computer program element comprising program instructions, which, when executed by one or more processors, cause the one or more processors to perform the method of any one of the above embodiments is provided.

[0028] According to various embodiments, a computer-readable medium comprising program instructions, which, when executed by one or more processors, cause the one or more processors to perform the method of any one of the above embodiments is provided. The computer-readable medium may include a non-transitory computer-readable medium.BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The invention will be better understood with reference to the detailed description when considered in conjunction with the non-limiting examples and the accompanying drawings, in which:- FIGS. 1 and 2 illustrate infrastructures of a system including a server for facilitating providing a recommendation to a merchant for the on-demand service according to various embodiments.- FIG. 3 illustrates a block diagram of a server for facilitating providing a recommendation to a merchant for the on-demand service according to various embodiments.- FIG. 4 illustrates a flowchart for a method for facilitating providing a recommendation to a merchant for the on-demand service according to various embodiments.FIG. 5 illustrates a data flow diagram of a system for facilitating providing a recommendation to a merchant for the on-demand service according to various embodiments.- FIG. 6 illustrates a data flow diagram of a system for facilitating providing a recommendation to a merchant for the on-demand service according to various embodiments.DETAILED DESCRIPTION

[0030] The following detailed description refers to the accompanying drawings that show, by way of illustration, specific details and embodiments in which the disclosure may be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the disclosure. Other embodiments may be utilized and structural, and logical changes may be made without departing from the scope of the disclosure. The various embodiments are not necessarily mutually exclusive, as some embodiments can be combined with one or more other embodiments to form new embodiments.

[0031] Embodiments described in the context of one of a server and a method are analogously valid for the other server and method. Similarly, embodiments described in the context of a server are analogously valid for a method, and vice-versa.

[0032] Features that are described in the context of an embodiment may correspondingly be applicable to the same or similar features in the other embodiments. Features that are described in the context of an embodiment may correspondingly be applicable to the other embodiments, even if not explicitly described in these other embodiments. Furthermore, additions and / or combinations and / or alternatives as described for a feature in the context of an embodiment may correspondingly be applicable to the same or similar' feature in the other embodiments.

[0033] In the context of various embodiments, the articles “a”, “an” and “the” as used with regard to a feature or element include a reference to one or more of the features or elements.

[0034] As used herein, the term “and / or” includes any and all combinations of one or more of the associated listed items.

[0035] Throughout the description, the term “module” may be understood as an application specific integrated circuit (ASIC), an electronic circuit, a combinational logic circuit, a field programmable gate array (FPGA), a processor which executes code, other suitable hardware components which provide the described functionality, or any combination thereof. The term of “module” may include a memory which stores code executed by the processor.

[0036] In the following, embodiments will be described in detail.

[0037] FIGS. 1 and 2 illustrate infrastructures of a system 200 including a server 100 for facilitating providing a recommendation to a merchant for the on-demand service according to various embodiments.

[0038] As shown in FIG. 1, the system 200 may include, but is not limited to, the server 100, a database system 140, a network 150, a first computing device 160 associated with the user 161 (also referred to as a “requester”), a plurality of second computing devices 170 (not shown) each associated with a plurality of delivery service providers 171 (also referred to as a “driver”, a “delivery partner” or a “delivery agent”), and a plurality of third computing devices 180 (not shown) each associated with a plurality of merchants 181 (also referred to as “item providers”) (for example, a “food provider” or a “restaurant”). In some embodiments, the user 161 may include a consumer (also referred to as an “eater”) for the on-demand service. For example, the user 161 may be the same as the consumer. As another example, the user 161 may be different from the consumer and use the on-demand service for the consumer.

[0039] In some embodiments, the on-demand service may be a service allowing the user 161 to fulfil the user’s demand via an immediate access to items and / or services. The user 161 mayrequest the on-demand service, such as a transport sendee (also referred to as a “ride-hailing sendee”) or an item delivery service, using a user interface presented on the first computing device 160. The user 161 may make an order for the on-demand service.

[0040] In some embodiments, the user 161 may use an application, for example, a mobile application, provided by the server 100. For example, the server 100 may be controlled and / or managed by an on-demand service platform provider. The application may be installed in the first computing device 160 associated with the user 161, to interact with the server 100.

[0041] In some embodiments, the plurality of delivery service providers 171 may use the application provided by the server 100. The application may be installed in the plurality of second computing devices 170 each associated with the plurality of delivery service providers 171, to interact with the server 100.

[0042] In some embodiments, the network 150 may include, but is not limited to, a Local Area Network (LAN), a Wide Area Network (WAN), a Global Area Network (GAN), or any combination thereof. The network 150 may provide a wireline communication, a wireless communication, or a combination of the wireline and wireless communication between the sc rvc i' 100 and the first computing device 160, between the server 100 and the plurality of second computing devices 170, and between the server 100 and the plurality of third computing devices 180. As shown in FIG. 2, the network 150 may provide the wireline communication, the wireless communication, or the combination of the wireline and wireless communication between the first computing device 160 and a second computing device 170a of the plurality of second computing devices 170.

[0043] In some embodiments, the first computing device 160 may be connectable to the server 100 via the network 150. In some embodiments, the first computing device 160 may be arranged in data or signal communication with the server 100 via the network 150. In some embodiments, the first computing device 160 may include, but is not limited to, at least one ofthe following: a mobile phone, a tablet computer, a laptop computer, a desktop computer, a head-mounted display and a smart watch. In some embodiments, the first computing device 160 may be associated with the user 161. For example, the first computing device 160 may belong to the user 161 who is the consumer. As another example, the first computing device 160 may belong to the user 161 requesting the delivery of the item to the consumer who is a recipient of the on-demand service. Although not shown, it may be appreciated that the system 200 may further include a plurality of first computing devices each associated with, for example, belonging to, a plurality of users.

[0044] In some embodiments, the first computing device 160 may include a location sensor. In some embodiments, the location sensor may communicate with at least one of a global positioning satellite (GPS) server, a network server, and a Wi-Fi server, to detect a location of the first computing device 160. In some embodiments, the first computing device 160 may generate information about the location of the first computing device 160.

[0045] In some embodiments, each of the plurality of second computing devices 170 may include a location sensor. In some embodiments, the location sensor may communicate with at least one of a global positioning satellite (GPS) server, a network server, and a Wi-Fi server, to detect a location of each of the plurality of second computing devices 170. In some embodiments, each of the plurality of second computing devices 170 may generate information about the location of each of the plurality of second computing devices 170.

[0046] In some embodiments, the server 100, for example, implemented by a server computer, may include a communication interface 110, a processor 120, and a memory 130 (as will be described with reference to FIG. 3).

[0047] In some embodiments, the server 100 may communicate with the first computing device 160 via the network 150. In some embodiments, the first computing device 160 may receive a request (hereinafter, referred to as an “order” or a “booking”) from the user 161 for the on-demand service. The first computing device 160 may send the order to the server 100 via the network 150. In some embodiments, the first computing device 160 may send the information about the location of the first computing device 160 to the server 100 via the network 150. The location of the first computing device 160 may be considered as a location of the user 161. In some embodiments, the location of the user 161 may be considered as a destination of the on-demand service. In some other embodiments, the first computing device 160 may send information about an address of the user 161, and the address of the user 161 may be considered as the destination of the on-demand service. In some other embodiments, the first computing device 160 may send information about an address of the recipient of the delivery, and the address of the recipient may be considered as the destination of the on-demand service.

[0048] In some embodiments, the system 200 may further include a database 141. In some embodiments, the database 141 may be a part of the database system 140 which may be external to the server 100. The server 100 may communicate with the database 141. In some other embodiments, although not shown, the database 141 may be implemented locally in the memory 130 of the server 100.

[0049] In some embodiments, the server 100 may communicate with the plurality of second computing devices 170 via the network 150. In some embodiments, the plurality of second computing devices 170 may be arranged in data or signal communication with the server 100 via the network 150. In some embodiments, the plurality of second computing devices 170 may include, but is not limited to, at least one of the following: a mobile phone, a tablet computer, a laptop computer, a desktop computer, a head-mounted display and a smart watch. The plurality of second computing devices 170 may be associated with the plurality of deliver}' service providers 171 respectively. For example, the plurality of second computing devices 170 may belong to the plurality of delivery service providers 171 respectively.

[0050] In some embodiments, the server 100 may receive the order from the first computing device 160. After the server 100 receives the order from the first computing device 160, the server 100 may allocate (assign) the order to a suitable delivery service provider 171a. In some embodiments, the second computing device 170a associated with the delivery service provider 171a may send information about a location of the second computing device 170a to the server 100 via the network 150. The location of the second computing device 170a may be considered as a location of the delivery service provider 171a. In some embodiments, the location of the delivery service provider 171a may be considered as a current location of the delivery service provider 171a, and may change while the delivery service provider 171a moves to the destination (en route to the destination). In some embodiments, the server 100 may provide the second computing device 170a with a map relating to a route from the current location of the second computing device 170a (which may be considered as the location of the delivery service provider 171a) to the destination for providing the on-demand service, en route to the destination. In some embodiments, the map may include an image object, for example, a pin, indicating the destination.

[0051] In some embodiments, the server 100 may communicate with the plurality of third computing devices 180 via the network 150. In some embodiments, the plurality of third computing devices 180 may be arranged in data or signal communication with the server 100 via the network 150. In some embodiments, the plurality of third computing devices 180 may include, but is not limited to, at least one of the following: a mobile phone, a tablet computer, a laptop computer, a desktop computer, a head-mounted display and a smart watch. The plurality of third computing devices 180 may be associated with the plurality of merchants 181 respectively. For example, the plurality of third computing devices 180 may belong to the plurality of merchants 181 respectively.

[0052] FIG. 3 illustrates a block diagram of a server 100 for facilitating providing a recommendation to a merchant for the on-demand service according to various embodiments.

[0053] As shown in FIG. 3, the server 100, for example, implemented by a server computer, may include a communication interface 110, a processor 120, and a memory 130.

[0054] In some embodiments, the memory 130 (also referred to as a “database”) may store input data and / or output data temporarily or permanently. In some embodiments, the memory 130 may be configured to store instructions. In some embodiments, the memory 130 may store program code which allows the server 100 to perform a method 300 (as will be described with reference to FIG. 4). In some embodiments, the program code may be embedded in a Software Development Kit (SDK). The memory 130 may include an internal memory of the server 100 and / or an external memory. The external memory may include, but is not limited to, an external storage medium, for example, a memory card, a flash drive, and a web storage.

[0055] In some embodiments, the communication interface 110 may allow one or more computing devices, including a first computing device 160, to communicate with the processor 120 of the server 100 via a network 150, as shown in FIG. 1. In some embodiments, as shown in FIG. 1, the first computing device 160 may belong to the user 161 who wants to make an order for the on-demand sendee. In some embodiments, the communication interface 110 may transmit signals to the first computing device 160, and / or receive signals from the first computing device 160 via the network 150.

[0056] In some embodiments, the communication interface 110 may allow a plurality of second computing devices 170 to communicate with the processor 120 of the server 100 via the network 150, as shown in FIGS. 1 and 2. As shown in FIGS. 1 and 2, each of the plurality of second computing devices 170 may belong to each of a plurality of delivery service providers 171 who may pick up an item from an item service provider 181a and deliver the item to the consumer (i.e. a destination) and / or who may transport the consumer to the destination. In someembodiments, the communication interface 110 may transmit signals to the plurality of second computing devices 170, and / or receive signals from the plurality of second computing devices 170, via the network 150.

[0057] In some embodiments, the communication interface 110 may allow a plurality of third computing devices 180 to communicate with the processor 120 of the server 100 via the network 150, as shown in FIG. 1. As shown in FIG. 1, each of the plurality of third computing devices 180 may belong to each of a plurality of merchants who may prepare an item, for example, food, for the order. In some embodiments, the communication interface 110 may transmit signals to the plurality of third computing devices 180, and / or receive signals from the plurality of third computing devices 180, via the network 150.

[0058] The processor 120 may include, but is not limited to, a microprocessor, an analogue circuit, a digital circuit, a mixed-signal circuit, a logic circuit, an integrated circuit, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), a Digital Signal Processor (DSP), a Field Programmable Gate Array (FPGA), an Application Specific Integrated Circuit (ASIC), or any combination thereof. Any other kind of implementation of the respective functions, which will be described below in further detail, may also be understood as the processor 120.

[0059] In some embodiments, the processor 120 may be connectable to the communication interface 110. In some embodiments, the processor 120 may be arranged in data or signal communication with the communication interface 110 to transmit / receive the signals.

[0060] In some embodiments, the communication interface 110 may receive a request (also referred to as an “order” or a “booking”) for the on-demand service from the first computing device 160 associated with the user 161 (also referred to as a “consumer 161”). In some embodiments, the processor 120 may receive the order for the on-demand service from the communication interface 110. In some embodiments, the processor 120 may allocate the order to the delivery sendee provider 171a among the plurality of the delivery service providers 171.In some embodiments, the processor 120 may control the second computing device 170a associated with the delivery sendee provider 171a to display a map relating to a route to from a current location of the second computing device 170a (which may be considered as the current location of the delivery service provider 171a) to a destination for providing the on-demand service, while the deliver}' service provider 171a moves to the destination (en route to the destination). In some embodiments, the destination may be a location of the first computing device 160 (which may be considered as the location of the consumer (i.e. the user 161)). In some other embodiments, the destination may be a location of a recipient (consumer) of the on-demand service that the requester (i.e. the user 161) has designated. In some embodiments, the map may include an image object indicating the destination. For example, the image object includes a pin (for example, a normal pin) indicating the destination.

[0061] There may be the plurality of merchants 181. In some embodiments, the processor 120 may group each merchant of the plurality of merchants 181 into a plurality ofclusters, to ensure benchmarking is relevant and meaningful. Specifically, in some embodiments, for each merchant of the plurality of merchants 181, the processor 120 may classify the merchant into at least one cluster of the plurality of clusters based on one or more criteria relating to the merchant. In some embodiments, the one or more criteria may include, but are not limited to, a geographical location of the merchant (for example, “city_id”), a geographical segmentation or a business circle level within a specific city (for example, “business_circle_index”), a size category of the merchant (for example, mini, small, medium, enterprise, etc.) based on a revenue and an average order value (for example, “segment”), and a main cuisine type offered in a menu (for example, “primary _cuisinc_id”). For example, a small merchant A serving a bubble tea as a main menu in a city and a small merchant B serving a bubble tea as a main menu in the same city may be grouped into the same cluster.

[0062] In some embodiments, the processor 120 may determine if the number of the two or more merchants belonging to the cluster is less than a threshold. If it is determined that the number of the two or more merchants belonging to the cluster is less than the threshold, the processor 120 may remove at least one criterion from the one or more criteria for re-classifying the plurality of merchants 181. Specifically, in some embodiments, the processor 120 may determine if the cluster size is below the threshold (for example, less than 7 transacting merchants). For example, the threshold may be determined by the on-demand service platform provider. For example, the threshold may be determined based on the geographical location of the merchant. It may be appreciated that the threshold may be updated based on an external environment. In some embodiments, if the cluster size is below the threshold, the processor 120 may use a broader criteria. Specifically, in some embodiments, the processor 120 may remove the at least one criterion (for example, “business_circle_index”) from the one or more criteria. In some embodiments, the processor 120 may re-classify each merchant of the plurality of merchants 181 into the plurality of clusters based on the updated criteria that the at least one criterion (for example, “business_circle_index”) is removed. In some embodiments, the processor 120 may determine if the cluster size is still below the threshold (for example, less than 7 transacting merchants). In some embodiments, if the cluster size is still below the threshold, the processor 120 may further simply the criteria. For example, the processor 120 may further simplify the criteria to “city_id” and “segment” only. In some embodiments, the processor 120 may output a clustering index for each merchant of the plurality of merchants 181 after classifying the plurality of merchants 181. In some embodiments, the clustering index may be an alphanumeric representation of the cluster. As an example, merchants belonging to the same clustering index may belong to the same cluster.

[0063] In some embodiments, the processor 120 may collect information about a plurality of features (also referred to as “attributes”) of two or more merchants belonging to the cluster thatthe merchant belongs to. In some embodiments, the processor 120 may collect the information about the plurality of features of the two or more merchants over a past predetermined period (for example, 2 weeks). In some embodiments, the processor 120 may gather various attributes for each merchant. In some embodiments, the plurality of features may include, but are not limited to, sales data, operational metrics, and menu features. In some embodiments, after collecting the information, in a step of a pre-processing, the processor 120 may aggregate the collected information about the plurality of features of the two or more merchants over the past predetermined period (for example, 2 weeks), to smoothen noisy features and signals, and balance data freshness and recency. In some embodiments, after the step of the pre-processing, in a step of a normalisation, the processor 120 may normalise the collected information about the plurality of features of the two or more merchants. Specifically, in some embodiments, the processor 120 may apply a standard z-scaling to the collected information, to standardise the collected information. This may be to make the collected information suitable for modelling. It may be appreciated that, in some other embodiments, the processor 120 may use other scaling technique, to standardise the collected information.

[0064] In some embodiments, the processor 120 may predict normalised sales based on the collected information, using a machine learning model. In some embodiments, the processor 120 may train the machine learning model, for example, a regression machine learning model, to predict the normalised sales based on the collected information. The machine learning model trained may be used to later calculate a Shapley value (from cooperative game theory). For the Shapley value to be accurate, a robust machine learning model which is efficient for large datasets and provides better prediction accuracy may need to be used.

[0065] In some embodiments, the processor 120 may calculate the Shapley value of each feature of the plurality of features for each merchant, based on the predicted normalised sales, using the machine learning model. In some embodiments, a negative Shapley value mayindicate that the corresponding feature may detract from a sales performance. In some embodiments, by performing the Shaplcy value analysis, the processor 120 may determine a contribution of each feature of the plurality of features to the predicted normalised sales. For example, the processor 120 may determine the contribution of each feature of the plurality of features based on the calculated Shapley value.

[0066] In some embodiments, the processor 120 may generate the recommendation based on the calculated Shapley value. In some embodiments, the processor 120 may generate the recommendation based on the determined contribution of each feature. In some embodiments, to identify opportunities, the processor 120 may focus on features with negative Shapley values, as these represent areas where the merchant lags behind competitors (peers). Specifically, in some embodiments, the processor 120 may extract at least one feature with a negative Shapley value from the plurality of features, as an adjustable feature. In some embodiments, after identifying the opportunities, the processor 120 may estimate a potential sales gain. In some embodiments, the processor 120 may estimate the potential sales gain by re-calculating the Shapley value using the adjustable feature and identifying a difference between the calculated Shapley value and the re-calculated Shapley value. Specifically, in some embodiments, to estimate the potential sales gain, the processor 120 may assume the merchant improves the adjustable feature to match the cluster mean (benchmark), and recalculate the Shapley value which may become zero if the adjustable feature matches the benchmark. In some embodiments, the difference between the calculated Shapely value (also referred to as a “current Shapley value”) and the re-calculated Shapley value (also referred to as a “benchmark Shaplcy value”) may represent a potential uplift. In some embodiments, the processor 120 may unnormalize the potential uplift (for example, using the saved “cluster_mean” and “cluster_stddev”) to estimate an actual percentage sales gain.

[0067] In some embodiments, the processor 120 may prioritise potential recommendations (also referred to as “potential actions’’). In some embodiments, if there arc two or more adjustable features, the processor 120 may rank the two or more adjustable features based on the corresponding potential sales gain, and select at least one adjustable feature from the two or more adjustable features based on the ranking, to generate the recommendation. Specifically, in some embodiments, the processor 120 may perform sorting, by ranking the potential recommendations based on a magnitude of the potential sales gain. In some embodiments, the processor 120 may perform the selection, by presenting top recommendations to the merchant, focusing on those with the highest expected impact (for example, the highest potential sales gain).

[0068] In some embodiments, the processor 120 may generate the recommendation using a Large Language Model (LLM), and provide the recommendations in clear and actionable language. In some embodiments, the processor 120 may provide the merchant’s current value, the benchmark and the estimated gain as an input to the LLM, and generate a dynamic recommendation message and / or content to be shown to the merchant.

[0069] As described above, the various embodiments may provide an action-driven insights recommendation system designed to assist the merchants in boosting their sales by providing prioritised and customised action recommendations. The system may leverage advanced machine learning models to analyse merchant data, benchmark performance against relevant competitors (peers), and identify the most impactful actions the merchant can take.

[0070] The various embodiments may have advantages over existing solutions. For example, unlike generic reports or onc-sizc-fits-all recommendations, the various embodiments may provide personalised and prioritised actions based on the merchant’s specific circumstances and comparison with the competitors (peers). The various embodiments may bridge a gapbetween data analytics and practical business decisions, enabling the merchants to act swiftly and effectively to improve their sales.

[0071] In addition, the various embodiments have the following advantages:• Data-Driven Prioritisation: Utilising the machine learning models (for example, LightGBM) and the Shaplcy values to predict sales and identify which features most significantly impact performance.• Customised Recommendations: Tailoring the recommendations based on individual merchant data and relevant peer benchmarks, ensuring the recommendations are highly relevant.• Actionable Insights: Translating complex data analytics into simple and clear actions that the merchants can implement without needing specialised expertise.• Time Efficiency: Saving merchants’ time by highlighting the most critical areas to focus on, eliminating the need to sift through extensive data.• Scalability: Offering a scalable solution that can serve a large number of merchants across different segments and locations.

[0072] FIG. 4 illustrates a flowchart for a method 300 for facilitating providing a recommendation to a merchant for the on-demand sendee according to various embodiments. According to various embodiments, the method 300 for facilitating providing the recommendation to the merchant for the on-demand sendee may be provided.

[0073] In some embodiments, the method 300 may include a step 301 of, for each merchant of a plurality of merchants, classifying a merchant into at least one cluster of a plurality of clusters based on one or more criteria relating to the merchant.

[0074] In some embodiments, the method 300 may include a step 302 of collecting information about a plurality of features of two or more merchants belonging to the cluster that the merchant belongs to.

[0075] In some embodiments, the method 300 may include a step 303 of predicting normalised sales based on the collected information about the plurality of features of the two or more merchants, using a machine learning model.

[0076] In some embodiments, the method 300 may include a step 304 of calculating a Shapley value of each feature of the plurality of features, based on the predicted normalised sales, using the machine learning model.

[0077] In some embodiments, the method 300 may include a step 305 of generating a recommendation for the merchant based on the calculated Shapley value. In some embodiments, the step 305 may include generating a recommendation message with LLM by providing LLM context of recommendation feature / information.

[0078] FIG. 5 illustrates a data flow diagram of a system 200 for facilitating providing a recommendation to a merchant for the on-demand service according to various embodiments.

[0079] In some embodiments, the various embodiments may operate through a multi-step methodology designed to process merchant data, identify key areas for improvement, and generate actionable recommendations prioritised by potential sales gain.

[0080] In some embodiments, a processor 120 of a server 100 (as described with reference to FIG. 3) may perform a step of a data collection 401. In some embodiments, the processor 120 may collect information relating to one or more criteria of a plurality of merchants 181. In some embodiments, the information relating to the one or more criteria may include, but is not limited to, information relating to a geographical location of the merchant, a geographical segmentation or a business circle level within a specific city, a size category of the merchant based on a revenue and an average order value, and a main cuisine type offered in a menu.

[0081] In some embodiments, the processor 120 may perform a step of merchant (also referred to as a “Mex”) clustering and a step of data pre-processing 402.

[0082] In some embodiments, in the step of the merchant clustering 402, the processor 120 may perform the merchant clustering for peers benchmarking. In some embodiments, the processor 120 may group each merchant of the plurality of merchants 181 into a plurality of clusters using the information collected in the step of the data collection 401, to ensure benchmarking is relevant and meaningful. Specifically, in some embodiments, for each merchant of the plurality of merchants 181, the processor 120 may classify the merchant into at least one cluster of the plurality of clusters based on the one or more criteria relating to the merchant. In some embodiments, the one or more criteria may include, but are not limited to, the geographical location of the merchant (for example, “city_id”), the geographical segmentation or the business circle level within the specific city (for example, “busincss_circlc_indcx”), the size category of the merchant (for example, mini, small, medium, enterprise, etc.) based on the revenue and the average order value (for example, “segment”), and the main cuisine type offered in the menu (for example, “primary _cuisine_id”).

[0083] In some embodiments, the processor 120 may determine if the cluster size is below the threshold (for example, less than 7 transacting merchants). In some embodiments, if the cluster size is below the threshold, the processor 120 may use a broader criteria. Specifically, in some embodiments, the processor 120 may remove the at least one criterion (for example, “business_circle_index”) from the one or more criteria. In some embodiments, the processor 120 may re-classify each merchant of the plurality of merchants 181 into the plurality of clusters based on the updated criteria that the at least one criterion (for example, “business_circle_index”) is removed. In some embodiments, the processor 120 may determine if the cluster size is still below the threshold (for example, less than 7 transacting merchants). In some embodiments, if the cluster size is still below the threshold, the processor 120 may further simply the criteria. For example, the processor 120 may further simplify the criteria to “city_id” and “segment” only. In some embodiments, the processor 120 may output a clusteringindex for each merchant of the plurality of merchants 181 after classifying the plurality of merchants 181. In some embodiments, the clustering index may be an alphanumeric representation of the cluster. As an example, merchants belonging to the same clustering index may belong to the same cluster.

[0084] In some embodiments, in the step of the data pre-processing 402, the processor 120 may gather various attributes for each merchant. In some embodiments, the plurality of features may include, but are not limited to, sales data, operational metrics, and menu features. For example, a list of merchant-level attribute and its description is as follows:• photo_rate - Percentage of menu items with photos• desc_rate - Percentage of menu items with descriptions• avg_rating - Average rating of all orders for the given merchant• has_hero_photo - Boolean indicating whether a banner / hero photo is available• has_listing_photo - Boolean indicating whether a listing photo is available• num_combo - Number of combo items in the menu• days_open - Number of days the merchant has been open• Fpt - Average food preparation time (in minutes)• Dwt - Average driver waiting time (in minutes)• mex_attr_cancel_rate - Percentage of orders cancelled due to merchant- attributable reasons• online_avail_rate - Percentage of time the merchant is available to receive orders (pauses reduce this rate)• item_availability_rate - Percentage of visible time when menu items are available • cnt_translated_language_100pct - Number of languages with full translation coverage for all menu items• total_ad_spend_usd - Total advertisement spend in USD• total_mfc_spend_usd -Total promotional campaign spend in USD

[0085] In some embodiments, after gathering the various attributes, for the pre-processing, the processor 120 may aggregate the collected information about the plurality of features of the two or more merchants over the past predetermined period (for example, 2 weeks), to smoothen noisy features and signals, and balance data freshness and recency. In some embodiments, after the pre-processing, the processor 120 may normalise the collected information about the plurality of features of the two or more merchants. Specifically, in some embodiments, the processor 120 may apply a standard z-scaling to the collected information, to standardise the collected information. This may be to make the collected information suitable for modelling. For example, the normalised collected information may be obtained by the following Equation (1):sales_normalised = (sales - cluster_mean) / cluster_stddev Equation (1)where the “sales” refers to an individual merchant’s sales (for example, last 2 weeks), the “cluster_mean” refers to an average sales of the merchant’s peers (in the cluster), and the “cluster_stddev” refers to a standard deviation in sales of the merchant’s peers (in the cluster). In some embodiments, the “cluster_mean” and the “cluster_stddev” may be used to normalise the sales. In some embodiments, the “cluster_mean” and “cluster_stddev” may be saved for a later unnormalisation. It may be appreciated that, in some other embodiments, the processor 120 may use other scaling technique, to standardise the collected information.

[0086] In some embodiments, the processor 120 may perform a step of a model training 403. In some embodiments, the processor 120 may train a machine learning model, for example, a regression machine learning model, to predict the normalised sales based on the collected information. The machine learning model trained may be used to later calculate a Shapleyvalue. For the Shapley value to be accurate, a robust machine learning model which is efficient for large datasets and provides better prediction accuracy may need to be used.

[0087] In some embodiments, the processor 120 may perform a step of a Shapley value analysis 404. In some embodiments, the processor 120 may calculate the Shapley value of each feature of the plurality of features for each merchant, based on the predicted normalised sales, using the machine learning model. In some embodiments, a negative Shapley value may indicate that the corresponding feature may detract from a sales performance. In some embodiments, by performing the Shapley value analysis, the processor 120 may determine a contribution of each feature of the plurality of features to the predicted normalised sales.

[0088] In some embodiments, the processor 120 may perform a step of generating a recommendation 405. In some embodiments, to identify opportunities, the processor 120 may focus on features with negative Shapley values, as these represent areas where the merchant lags behind competitors (peers). In some embodiments, to estimate the potential sales gain, the processor 120 may assume the merchant improves the adjustable feature to match the cluster mean (benchmark), and re-calculate the Shapley value which may become zero if the adjustable feature matches the benchmark. In some embodiments, the difference between the calculated Shapely value and the re-calculated Shapley value may represent a potential uplift. In some embodiments, the processor 120 may unnormalize the potential uplift (for example, using the saved “cluster_mean” and “cluster_stddev”) to estimate an actual percentage sales gain. For example, the calculation of the potential uplift may be as follows:• Feature: photo_rate (percentage of menu items with photos)• Merchant’s Value: 70%• Benchmark (Cluster Mean): 90%• Current Shapley Value: -0.5• Potential Uplift:New Shaplcy Value if the merchant reaches the benchmark: 0- Uplift: 0 - (-0.5) = 0.5- Unnormalise 0.5 to estimate actual sales gain (for example, 10% increase in sales).

[0089] In some embodiments, in the step of generating the recommendation 405, the processor 120 may prioritise the potential recommendations. In some embodiments, the processor 120 may perform sorting, by ranking the potential recommendations based on a magnitude of the potential sales gain. In some embodiments, the processor 120 may perform the selection, by presenting top recommendations to the merchant, focusing on those with the highest expected impact (for example, the highest potential sales gain).

[0090] In some embodiments, in the step of generating the recommendation 405, the processor 120 may provide an actionable recommendation. In some embodiments, the processor 120 may generate the recommendation using a Large Language Model (LLM), and provide the recommendations in clear and actionable language. In some embodiments, the processor 120 may provide the merchant’s current value, the benchmark and the estimated gain as an input to the LLM, and generate a dynamic recommendation message and / or content to be shown to the merchant. Examples of the actionable recommendation are as follows:• Photo Coverage: “Your menu photo coverage is 70%, which is below the peer average of 90%. Increasing your photo coverage to match peers could boost orders by up to 10%. ”• Average Rating: “Your average customer rating is 4.2, while peers average 4.7.Improving your customer service could increase your sales by approximately 6%.” • Operational Hours: “You are currently open 1 day a week, whereas peers are open 4 days. Expanding your operational days could boost sales by up to 30%.”

[0091] In some embodiments, the processor 120 may perform a step of displaying a recommendation 406. In some embodiments, the processor 120 may perform an integration ofthe recommendation with an on-demand sendee platform. For example, the processor 120 may embed the recommendation system within the merchant’ s dashboard on the on-demand service platform, and provide notifications or alerts when new recommendations are available. In some embodiments, the processor 120 may provide a user interface. For example, the processor 120 may provide a dashboard view which may display prioritised recommendations with visual indicators of the potential gain. As an example, the processor 120 may provide a detail view which may allow the merchants to click on a recommendation to view more details, including data visualisations and steps to implement the action. In some embodiments, the processor 120 may provide an action enablement service, for example, including quick action buttons (for example, CTA (Call-to-Action)) or deep-links to facilitate immediate implementation (for example, upload menu photos, adjust operational hours, etc.).

[0092] In some embodiments, a technical architecture of the server 100 may include the following components:• Data Layer:- Data Sources: Collecting sales data along with merchant activities, transaction records, and customer interactions.- Data Storage: Using robust and scalable database like presto to store and manage merchant data.• Analytics Engine:- Model Training Module: Automating the training of the LightGBM model periodically to incorporate the latest data.- Shapley Value Calculator: Implementing algorithms to compute Shapley values efficiently.• Recommendation Engine:- Rules Engine: Applying a business logic to generate and format recommendations.Prioritisation Module: Ranking the recommendations based on estimated gains.• Presentation Layer:- API Services: Providing interfaces for the merchant dashboard to retrieve recommendations.- User Interface: Designing an intuitive and user-friendly screens for displaying insights and facilitating actions.

[0093] In some embodiments, in a step of feedback and performance tracking 407, the processor 120 may perform a maintenance and continuous improvement. In some embodiments, the processor 120 may provide a feedback loop, by allowing the merchants to provide feedback on recommendations (for example, relevance, effectiveness, etc.), and using the feedback to refine the machine learning model and improve future recommendations. In some embodiments, the processor 120 may perform a performance monitoring, by tracking the impact of implemented actions on merchant sales, and updating benchmarks and the machine learning model accordingly to maintain accuracy.

[0094] In some embodiments, the processor 120 may consider security and privacy. In some embodiments, the processor 120 may perform a data protection, by ensuring a compliance with data protection regulations (for example, GDPR, PDPA), and implementing robust security measures to protect sensitive merchant data. In some embodiments, the processor 120 may perform a consent management, by obtaining necessary permissions from the merchants for data usage, and providing transparency on how data is used to generate the recommendations.

[0095] In some embodiments, the processor 120 may consider scalability and localisation. In some embodiments, for the scalability, the processor 120 may design the system to handle a large number of merchants across different regions without performance degradation. In some embodiments, for the localisation, the processor 120 may adapt the recommendations to localmarket conditions and cuiturai contexts, and support multipic languages for broader accessibility.

[0096] FIG. 6 illustrates a data flow diagram of a system 200 for facilitating providing a recommendation to a merchant for the on-demand sendee according to various embodiments. FIG. 6 shows a high-level solution framework. The system 200 may operate through the following components, as illustrated in FIG. 6.

[0097] In some embodiments, the system 200 may include a data source (also referred to as a “Mex datastore”) 501. In some embodiments, the data source 501 may be a centralised database including all merchant-related attributes (for example, historical engagement metrics, features, CTR (Click-Through Rate), CVR (Conversion Rate), etc.) at a merchant ID level. In some embodiments, the data source 501 may serve as a foundational dataset for all recommendation engines and delivery optimisation algorithms.

[0098] In some embodiments, the system 200 may further include a recommendation engine 502. In some embodiments, the recommendation engine 502 may be a suite of machine learning or rules-based models (for example, Competitor based model, Model B / C / D, etc.) designed to generate personalised recommendations based on various data dimensions (for example, competitor performance, customer reviews, behaviour, etc.). In some embodiments, the competitor based recommendation model may be just one of several models that goes as an input to a merchant recommendation delivery system 503.

[0099] In some embodiments, the system 200 may further include the merchant recommendation delivery system 503. In some embodiments, the merchant recommendation delivery system 503 may be a core decision-making layer that determines the following:• Recommendation Model Selection: Implementing algorithms like Multi-Armed Bandit (MAB) / Thompson sampling strategies to dynamically select the optimal recommendation engine for each merchant.Timing Optimisation: Leveraging historical engagement data to predict the best time / day for recommendation delivery.• Channel Optimization: Analysing merchant preferences and past interactions to identify the most effective delivery channel.

[0100] In some embodiments, the system 200 may further include an LLM-generated message evaluation module 504. In some embodiments, the LLM-generated message evaluation module 504 may integrate a Large Language Model (LLM) to evaluate and finetune the generated recommendation messages for clarity, relevance, and increased engagement.

[0101] In some embodiments, the system 200 may further include delivery channels 505. In some embodiments, the delivery channels 505 may include multiple communication channels, including, but not limited to, a chatbot, an insights page (nudges), an application inbox notification, and an email.

[0102] In the following, various examples of this disclosure are illustrated:

[0103] Example 1 is a server for facilitating providing a recommendation to a merchant for an on-demand service, the server comprising: a memory configured to store instructions; and a processor for executing the stored instructions and configured to: for each merchant of a plurality of merchants, classify the merchant into at least one cluster of a plurality of clusters based on one or more criteria relating to the merchant; collect information about a plurality of features of two or more merchants belonging to the cluster that the merchant belongs to; predict normalised sales based on the collected information about the plurality of features of the two or more merchants, using a machine learning model; calculate a Shaplcy value of each feature of the plurality of features, based on the predicted normalised sales, using the machine learning model; and generate the recommendation based on the calculated Shapley value.

[0104] Example 2 is the server according to Example 1, wherein the processor is further configured to: determine if a number of the two or more merchants belonging to the cluster isless than a threshold; and if it is determined that the number of the two or more merchants belonging to the cluster is less than the threshold, remove at least one criterion from the one or more criteria for re-classifying the merchant.

[0105] Example 3 is the server according to Example 1 or Example 2, wherein the processor is further configured to: collect the information about the plurality of features of the two or more merchants over a past predetermined period; and aggregate the collected information about the plurality of features of the two or more merchants over the past predetermined period.

[0106] Example 4 is the server according to any one of Examples 1 to 3, wherein the processor is further configured to: normalise the collected information about the plurality of features of the two or more merchants.

[0107] Example 5 is the server according to Example 4, wherein the processor is further configured to: apply a standard z- scaling to the collected information about the plurality of features of the two or more merchants, to standardise the collected information about the plurality of features of the two or more merchants.

[0108] Example 6 is the server according to any one of Examples 1 to 5, wherein the processor is further configured to: train the machine learning model to predict the normalised sales based on the collected information about the plurality of features of the two or more merchants.

[0109] Example 7 is the server according to any one of Examples 1 to 6, wherein the processor is further configured to: determine a contribution of each feature of the plurality of features to the predicted normalised sales based on the calculated Shapley value; and generate the recommendation based on the determined contribution of each feature.

[0110] Example 8 is the server according to any one of Examples 1 to 7, wherein the processor is further configured to: extract at least one feature with a negative Shapley value from the plurality of features, as an adjustable feature.

[0111] Example 9 is the server according to Example 8, wherein the processor is further configured to: estimate a potential sales gain by re-calculating a Shaplcy value using the adjustable feature and identifying a difference between the calculated Shapley value and the re-calculated Shapley value; and if there are two or more adjustable features, rank the two or more adjustable features based on the corresponding potential sales gain, and select at least one adjustable feature from the two or more adjustable features based on the ranking, to generate the recommendation.

[0112] Example 10 is the server according to any one of Examples 1 to 9, wherein the processor is further configured to: generate the recommendation using a Large Language Model (LLM).

[0113] Example 11 is a method for facilitating providing a recommendation to a merchant for an on-demand service, the method comprising: for each merchant of a plurality of merchants, classifying the merchant into at least one cluster of a plurality of clusters based on one or more criteria relating to the merchant; collecting information about a plurality of features of two or more merchants belonging to the cluster that the merchant belongs to; predicting normalised sales based on the collected information about the plurality of features of the two or more merchants, using a machine learning model; calculating a Shapley value of each feature of the plurality of features, based on the predicted normalised sales, using the machine learning model; and generating the recommendation based on the calculated Shapley value.

[0114] Example 12 is the method according to Example 11, further comprising: determining if a number of the two or more merchants belonging to the cluster is less than a threshold; and if it is determined that the number of the two or more merchants belonging to the cluster is less than the threshold, removing at least one criterion from the one or more criteria for reclassifying the merchant.

[0115] Example 13 is the method according to Example 11 or Example 12, further comprising: collecting the information about the plurality of features of the two or more merchants over a past predetermined period; and aggregating the collected information about the plurality of features of the two or more merchants over the past predetermined period.

[0116] Example 14 is the method according to any one of Examples 11 to 13, further comprising: normalising the collected information about the plurality of features of the two or more merchants.

[0117] Example 15 is the method according to Example 14, further comprising: applying a standard z-scaling to the collected information about the plurality of features of the two or more merchants, to standardise the collected information about the plurality of features of the two or more merchants.

[0118] Example 16 is the method according to any one of Examples 11 to 15, further comprising: training the machine learning model to predict the normalised sales based on the collected information about the plurality of features of the two or more merchants.

[0119] Example 17 is the method according to any one of Examples 11 to 16, further comprising: determining a contribution of each feature of the plurality of features to the predicted normalised sales based on the calculated Shapley value; and generating the recommendation based on the determined contribution of each feature.

[0120] Example 18 is the method according to any one of Examples 11 to 17, further comprising: extracting at least one feature with a negative Shapley value from the plurality of features, as an adjustable feature.

[0121] Example 19 is the method according to Example 18, further comprising: estimating a potential sales gain by re-calculating a Shapley value using the adjustable feature and identifying a difference between the calculated Shapley value and the re-calculated Shapley value; and if there are two or more adjustable features, ranking the two or more adjustablefeatures based on the corresponding potential sales gain, and selecting at least one adjustable feature from the two or more adjustable features based on the ranking, to generate the recommendation.

[0122] Example 20 is the method according to any one of Examples 11 to 19, further comprising: generating the recommendation using a Large Language Model (LLM).

[0123] While the disclosure has been particularly shown and described with reference to specific embodiments, it should be understood by those skilled in the art that various changes in form and detail may be made therein without departing from the spirit and scope of the invention as defined by the appended claims. The scope of the invention is thus indicated by the appended claims and all changes which come within the meaning and range of equivalency of the claims arc therefore intended to be embraced.

Claims

CLAIMS1. A server for facilitating providing a recommendation to a merchant for an on-demand service, the server comprising:a memory configured to store instructions; anda processor for executing the stored instructions and configured to: for each merchant of a plurality of merchants,classify the merchant into at least one cluster of a plurality of clusters based on one or more criteria relating to the merchant;collect information about a plurality of features of two or more merchants belonging to the cluster that the merchant belongs to;predict normalised sales based on the collected information about the plurality of features of the two or more merchants, using a machine learning model;calculate a Shapley value of each feature of the plurality of features, based on the predicted normalised sales, using the machine learning model; andgenerate the recommendation based on the calculated Shapley value.

2. The server according to claim 1, wherein the processor is further configured to:determine if a number of the two or more merchants belonging to the cluster is less than a threshold; andif it is determined that the number of the two or more merchants belonging to the cluster is less than the threshold, remove at least one criterion from the one or more criteria for reclassifying the merchant.

3. The server according to claim 1, wherein the processor is further configured to:collect the information about the plurality of features of the two or more merchants over a past predetermined period; andaggregate the collected information about the plurality of features of the two or more merchants over the past predetermined period.

4. The server according to claim 1, wherein the processor is further configured to: normalise the collected information about the plurality of features of the two or more merchants.

5. The server according to claim 4, wherein the processor is further configured to: apply a standard z-scaling to the collected information about the plurality of features of the two or more merchants, to standardise the collected information about the plurality of features of the two or more merchants.

6. The server according to claim 1, wherein the processor is further configured to: train the machine learning model to predict the normalised sales based on the collected information about the plurality of features of the two or more merchants.

7. The server according to claim 1, wherein the processor is further configured to:determine a contribution of each feature of the plurality of features to the predicted normalised sales based on the calculated Shapley value; andgenerate the recommendation based on the determined contribution of each feature.

8. The server according to claim 1, wherein the processor is further configured to: extract at least one feature with a negative Shapley value from the plurality of features, as an adjustable feature.

9. The server according to claim 8, wherein the processor is further configured to: estimate a potential sales gain by re-calculating a Shapley value using the adjustable feature and identifying a difference between the calculated Shapley value and the re-calculated Shapley value; andif there are two or more adjustable features, rank the two or more adjustable features based on the corresponding potential sales gain, and select at least one adjustable feature from the two or more adjustable features based on the ranking, to generate the recommendation.

10. The server according to claim 1 , wherein the processor is further configured to: generate the recommendation using a Large Language Model (LLM).

11. A method for facilitating providing a recommendation to a merchant for an on-demand service, the method comprising: for each merchant of a plurality of merchants, classifying the merchant into at least one cluster of a plurality of clusters based on one or more criteria relating to the merchant;collecting information about a plurality of features of two or more merchants belonging to the cluster that the merchant belongs to;predicting normalised sales based on the collected information about the plurality of features of the two or more merchants, using a machine learning model;calculating a Shapley value of each feature of the plurality of features, based on the predicted normalised sales, using the machine learning model; andgenerating the recommendation based on the calculated Shapley value.

12. The method according to claim 11, further comprising:determining if a number of the two or more merchants belonging to the cluster is less than a threshold; andif it is determined that the number of the two or more merchants belonging to the cluster is less than the threshold, removing at least one criterion from the one or more criteria for reclassifying the merchant.

13. The method according to claim 11, further comprising:collecting the information about the plurality of features of the two or more merchants over a past predetermined period; andaggregating the collected information about the plurality of features of the two or more merchants over the past predetermined period.

14. The method according to claim 11, further comprising: normalising the collected information about the plurality of features of the two or more merchants.

15. The method according to claim 14, further comprising: applying a standard z-scaling to the collected information about the plurality of features of the two or more merchants, to standardise the collected information about the plurality of features of the two or more merchants.

16. The method according to claim 11, further comprising: training the machine learning model to predict the normalised sales based on the collected information about the plurality of features of the two or more merchants.

17. The method according to claim 11, further comprising:determining a contribution of each feature of the pluraiity of features to the predicted normalised sales based on the calculated Shaplcy value; andgenerating the recommendation based on the determined contribution of each feature.

18. The method according to claim 11, further comprising: extracting at least one feature with a negative Shapley value from the plurality of features, as an adjustable feature.

19. The method according to claim 18, further comprising:estimating a potential sales gain by re-calculating a Shapley value using the adjustable feature and identifying a difference between the calculated Shapley value and the re-calculated Shaplcy value; andif there are two or more adjustable features, ranking the two or more adjustable features based on the corresponding potential sales gain, and selecting at least one adjustable feature from the two or more adjustable features based on the ranking, to generate the recommendation.

20. The method according to claim 11, further comprising: generating the recommendation using a Large Language Model (LLM).

Images