ML for Supply Chain Planning: As ML can be employed for predicting the future sales, it can also be utilized for maintaining hassle-free supply chain management despite involvement of several uncertain features. Moreover, it can also be used for optimized route planning for delivery of goods or warehouse maintenance. A route suggested by ML algorithms ensures to efficiently optimize cost, time, and carbon emission in comparison to humans.
ML for price Optimization: ML algorithms can be employed for predicting the optimized price of products considering several factors like amount of discount, product type, competing retailers, and time dimension. Usage of ML in price optimization yields accurate predictions over traditional methods of price optimization. It can also be used for revenue forecasting during a particular month, quarter, or financial year [25].
3.3.2 Use Cases
As discussed earlier, ML has been employed in several leading retail industries to sustain and excel in this cut-throat competing business world. Traditionally, managers would have predicted sales based on various factors like brand quality, promotion, and discount. Managers used to implement a series of regressions to predict the sales volume. The efficiency of such an approach heavily relied on the capability of the human brain. Traditional methods even led to inefficient forecasting. This inefficiency has been completely handled by incorporating ML approaches. In this subsection, authors discuss few popular use cases of ML employment in retail industries.
According to [26], a ML model is devised to predict sales in response to promotion by a multinational retailer. An efficient model would enable to garner a huge leap in sales. Here, the retail company wanted to have an idea about the strongly and weakly performing products in the store. In the model, the company used several variables like discount, promotion duration, size of promotional advertisement, placement of products, and seasonality, among others. It was observed in [26] that a traditional method which involved several data analysts and a series of linear regression models predicted the results with 30% to 35% error rate. This error rate was brought down in the first attempt to 24% using ML model, and the error rate is expected to further reduce over time. Thus, integration of ML approaches in prediction models provided exciting and attractive results. It helped to curb the cost involved in generous promotions and maintaining inventory in the warehouse. Using the similar predictive model, Target Corp. also observed the growth of 15%–30% in revenue.
A renowned retailer Walmart has also incorporated technologies to understand customers’ needs and act accordingly. The company employed facial recognition software to understand the experience level (frustration, happiness, and satisfaction) during checkout. It also triggers an alert for customer representatives to approach frustrated customers in order to provide better customer service. Usage of this facial recognition model eliminates the need of maintaining expensive and appropriate staff for providing enhanced customer service.
Amazon has been proudly employing ML for predictive data analytics to enhance its sales. Amazon has garnered its outstanding benefit for demand prediction in business management [27]. It has also filed a patent for the process of its anticipatory shipping that predicts sales of a product in a particular region or city. Amazon uses this information to store the targeted products in nearest warehouses. It is also planning to deliver the product to the customers using drones in minimum time thus excelling the experience of shopping.
Authors in [28] have presented the implementation of predictive analytics in the retail banking sector. Here, authors claim that predictive analytics through traditional tools necessitates a specialized skill in statistics and mathematics. However, the same can be performed much easily in R, a language that includes around 4,000 algorithms of ML ranging from basic regression model to advanced model. In the banking sector, predictive analytics can be used to estimate churning rate and product propensity.
3.3.3 Limitations and Challenges
ML has observed widespread deployment in various domains including retail industry. In the retail industry also, it has been implemented for numerous purposes as discussed earlier. Despite its widespread deployment, it has some limitations and challenges. The major challenge is handling an ocean of data from diverse sources involving structured, semi-structured, and even unstructured data. This huge data collected from various sources is generally of poor quality, and therefore, efficient data cleaning methods need to be used to infer meaningful data [17]. Thereafter, it also has the challenge of maintaining stringent privacy and security policies for this huge data. It also has a challenge of acquiring trained and competent professionals who have vision of the future data requirement so as companies are able to draw useful insights from historical data.
3.4 Proposed Model
In this section, authors propose a model for predictive data analytics in retail industries using ML approaches. Ahead of proposing the model, authors attempt to thoroughly understand the requirements of retailers. It is understood that retailers have various queries in mind which need to be addressed by an efficient model. Some of these queries are as follows:
What is the probability of a person who is predicted to have online purchase behavior truly purchases online?
Which segment of customers the retailer should focus on?
Which are the geographical regions for online and offline channels?
A detailed understanding of the various queries of retailers enables devising an efficient predictive model. Here, authors aim to devise a model that provides various functions. Some of these functions are illustrated in Figure 3.2.
For instance, the proposed model can be used to estimate and forecast the sales of a particular product for a particular region. It can be performed at various levels of abstraction as per retailer’s choice and requirement. The proposed model aims to find the prospective buyers for a product even with very little probability of purchase. As it is observed, if a model targets more customers, then it may involve some additional costs but will not miss any probable buyer. Authors aim to not miss any probable customer as it may result in loss of some potential customers. The proposed model also attempts to predict the likelihood of a customer purchasing a particular product. This helps in targeting the prospective customers thus yielding an increase in revenue.
Figure 3.2 Illustration of major functions of predictive data analytics.
The proposed model collects data from various sources like social media, history data, and transaction details. This data from diverse sources is in disparate forms and thus needs to be cleaned during preprocessing. Thus, cleaned data from various sources is integrated, which is used for training the predictive model. The accuracy of model is largely dependent upon the size of training dataset. The basic structure of proposed model is represented in Figure 3.3.
As represented in Figure 3.3, the data integration is followed by algorithm selection for predictive model. There are several related algorithms like regression, boosting, or bagging, to name a few. Regression algorithms are basic algorithms for any predictive model. Boosting algorithms trains a model in a sequential and gradual manner. These algorithms perform both classification and regression. Boosting algorithms basically aim to identify weak learners which further can be improvised so as it turns to be a strong learner. Gradient boosting and AdaBoost are the two popularly used boosting algorithms. These two boosting algorithms basically differ in identification of weak learners. Weak learners are identified based on error rate. Error rate depends on the parameters to be optimized. For