Research and Development of ML based prediction for human cognitive behavior of sales

Project Title

Research and Development of ML based prediction for human cognitive behavior of sales

Project Area of Specialization Artificial IntelligenceProject Summary

In urban areas, people are more frequently travelling to different places called the Point of Interest (POI) to buy, sell or to entertain themselves. In the research, the passenger flow has been a key collectable information in Pakistan that is being missed due no automation trip information inside a city. This information can be used to apply any analytics to prove the reasons behind the passenger flow in a certain time. We consider that coming era is the key research topic among the researchers analyzing people choice to estimate their cognitive nature. The behaviour of the people to travel towards a some specific place can be known by their cognitive data. In most cases a greater number of people travel to a place while other remain vacant. In case of absence of Automated Fare Collection data, the current statistics on the purpose of travelling and taking a drop-off possibly cause a surge on a route remain a questionnaire. This phenomenon causes surge in the traffic to a point and become difficult for transportation experts in driving the new rules to adjust the surge. Analysing the people choices by qualitative survey combined with their geolocations can help to develop a knowledge-based system that recommend them towards the potential buyers in their nearest store where the desired item is available. This will automatically contribute to reduce the item searching from an unknown item list of stores. The aim of our research is to collect the choices of people by their neurological attentions towards the prospect item and develop a model for the behaviour of customers towards a point of interest (POI) for purchasing an item based on its features. Outcomes of our model will sufficiently be able to predict the purchase pattern resulting appropriate products to be on sales. As a product, our contribution will directly improve the online sales of items across the stores through their nearby reducing the traffic surge in Karachi, Pakistan. For the experiment, we have collected data from poorly congested markets like Liaqatabad, Tariq Road and Hydri along with the geolocation of the POI, Store Item list and the Choices of people by Neurological Change in the EEG waves towards an item as their choice.

Project Objectives Project Objectives

• Research Objectives

1. To collect data of people choice and make an analysis of the items.
2. To reduce the waiting time of passengers to provide them with an optimized route to their destination POI.
3. To understand human cognitive behavior for sales prediction in Pakistan

• Academic Objectives

1. To work with machine learning algorithms using real-world data
2. To learn how to plan, organize and execute a project
3. To publish research article in an international conference/ journal.

• Commercial Objectives

1. To design a system that can be imported into an application or a web-based portal.
2. To provide a solution for an optimized route of public transport to the transportation authorities for improvising the urban mobility and economic infrastructure.

Project Implementation Method

As this is an ML-based research project so in the first phase extensive literature review will be done to conduct the research further. The conclusive research methodology aims to provide a well-defined structure that helps in formulating and solving the hypotheses and giving appropriate results. The results are generic which will help the outside world. There is an involvement of the buyers as well as their cognitive behaviour while buying an item is analyzed, with the features grabbing their attention towards an item on display. While doing the descriptive research, a questionnaire is generated to get feedback, which will further determine the facts

Data and Requirement Gathering 

Identification of the relevant data, data fields and data sources will be done in order to acquire the data set which will be taken through a process of pruning, balancing and pre-processing according to the models on which it will be implemented. 

The clothing data set will be extracted organically from the websites of clothing brands currently popular in Pakistan and people's choice data will be collected from a generated questionnaire including the most revenant questions vital for this research. 

Data collection from four major stores:
Khaadi 
GulAhmed
Sana Safinaz
Sapphire 

Questionnaire
The generated questionnaire will include relevant questions about the preferences of people between the age of 18-50 about their preferred brand, their budget, their most visited store, etc. 

Hardware
The hardware equipment "EMOTIV Insight 2.0 – 5 Channel Mobile Brainwear" will be used to analyse inferences people have regarding different features of a clothing item. The features on the basis of which they make a purchase and what they seek in an item in order for it to be an easy purchase. 

Project Design Analysis 

After the collection and processing of data, it will be much needed to extensively study the ML algorithms and models which are going to be used. The data will also be shaped in the form of input for those algorithms. 

The futher structure design of the project will be made in this phase to check if any changes are necessary.

Implementation 

Machine learning models such as KNN, TensorFlow, and LSTM will be implemented on the collected datasets and to check their accuracy their results will be analyzed.  

Testing 

After getting the desired accuracy of the models, a new testing dataset will be given to the models to verify the results and point out any potential errors.

 NOTE:  The cost of equipment used, "Emotive insight - 5 Channel Mobile Brainwear" is mentioned as Rs. 70,000 in the item details because of the limitation but it actually costs $499 i.e Rs. 92,664 and will cost around Rs. 95,000 with the delivery cost to Pakistan. So the additional price, i.e. Rs. 25,000, will be contributed by the students themselves. The equipment link has also been provided below.

Equipment Link: https://www.emotiv.com/product/emotiv-insight-5-channel-mobile-brainwear/

Benefits of the Project Project Benefits:

1. This system will increase the business sales of any clothing brand.
2. This system will cater both online and physical buyers altogether.
3. This system will provide an optimize route on basis of their desired item.
4. Employing this system means fulfilling the demand of clothing brands to increase their sales and provide an optimize route to both physical and online customers towards the optimal store holding the clothing item in under their price range.
5. In near future, this system will come into the form of a website and an application so that people may have full advantage of it.

Technical Details of Final Deliverable

The research on human cognitive behaviour of sales has the characteristics to provide a recommendation system which will help the user to find their desired clothing product available in their nearest store. With the help of analysis done on the preference of the user, the system will predict an item which is most likely to be sold in a store on the basis of its features. Store owners/ branch owners will be able to know what features attract their customers more to buy a certain clothing item and manage their inventory accordingly which will help them increase sales.

Final Deliverable of the Project HW/SW integrated systemCore Industry ITOther Industries Transportation Core Technology Big DataOther Technologies Artificial Intelligence(AI)Sustainable Development Goals Industry, Innovation and Infrastructure, Sustainable Cities and Communities, Climate ActionRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	80000
EMOTIV Insight 2.0 – 5 Channel Mobile Brainwear	Equipment	1	70000	70000
Arduino UNO Starter Kit	Miscellaneous	1	6000	6000
Printing, Report Making	Miscellaneous	1	2500	2500
Patch Cord for EEG	Miscellaneous	1	1500	1500