467 AI based management of Food Wastage Lakshit Samaa, Aaisha Makkarb, Polemoni Prokshithac, Bhav Kirti Sharmac, Devansh Dhaloriac a Deakin University, Waurn Ponds, Australia, b Thapar Institute of Engineering & Technology, Patiala, India c CSE Dept Chandigarh University, Chandigarh, India Abstract The problem we see here is of how food around us is wasted, even after having enough food to feed everyone, there's still hunger around us. The solution we are focusing on is not of the field, but to stop the wastage happening around us. The proposed scheme is to build an outline which contains general information to tackle similar kinds of problems. Just by taking a few initiatives we were actually able to reduce the food wastage in our hostel messes. Keywords 1 Food Wastage, Hunger Management ISIC‟21International Semantic Intelligence Conference, February EMAIL:lakshit.sama@gmail.com(a),aaisha.makkar@thapa r.edu(b),polemoninikky6@gmail.com(c,1),bhavkirtis@gm ail.com (c,2), devanshdhaloria@gmail.com(c,3) 2020 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0). CEUR Workshop Proceedings (CEUR-WS.org) 468 1. Introduction that are to be considered for the present study and to obtain effective and efficient prediction results which help us to implement in real time Use only styles embedded in the document. so as to decrease the food wastage [1]. These While we sit here tranquilly with our belly‟s strategies do not include the development and plenary, some might be out there without even improvement of existing structures but inform one grain of victuals. Now we can‟t solve the people to promote the reduction, recycling and quandary with some impulse of activity. What recycling of solid waste generated [1]. The we require to do is to act locally and make the analysis of garbage generation could be ones around us cognizant of the genuine accomplished with the help of data in the form quandary. According to some studies [7], over of a time series, which contains the quantity of 820 million people are suffering from garbage being produced [1]. starvation even after enough aliment for them A project by the students of IIT Mandi for on the planet. The quandary is not only providing long term solutions for reducing hunger, the victuals wasted engenders food wastage [2]. The techniques they adopted greenhouse gases. And a plethora of aliment for the same is: betokens an abundance of greenhouse gases which in some way affects the environment we  Determining the size of the food is live in. Around 3.3 billion Carbon dioxides are trashed engendered annually by the amplitude of  Exploring why the food is wasted victuals wastage ecumenically. Now after the  Using new techniques like artificial past incidents like amazon rain-forest fire and intelligence on accurate data [2]. Australian forest fire, the impact on the ecosystem and the environment is going to be After the visualization of the data, many more consequential than ever and that will just similarities are found between the two messes, be a slow doom for our planet. Even after all for example wastage of the same type of food we endeavor, we are not going to solve the on the same days and at the same time. By quandary entirely and that‟s because of the maintaining different waste bins for the mess variety of quandaries all around the world and labelled for different types of food wastes. that is because countries like Africa are facing Asking people like chefs, mess staff, college the scarcity of pabulum because of their lack staff members, students. The rental divisions of infrastructure and the lack of technology. in which they calculate the number of cooks, Not only the aliment wasted has an impact on determine the behavior of each cook and the environment but withal the economy of measure the waste generated on each food countries. For example, annual pabulum menu [2]. An unequal distribution in which all wastage can be summed up to 1 trillion US of the above points are recorded but without dollars. If we visually examine the topic, the the presence of a group. In this way they have quandaries caused by pabulum is hunger and used an Arduino-connected sensor so that the gases engendered due to victuals wastage every plate passing through the section is can be solved by taking certain parameters. calculated and hence the daily data related to hostel mess [2]. Data mining and decision trees are used to manage food. Data mining 2. Literature Survey basically collects all the data and analyzes 2.1. Related Work high-value data to find reasonable patterns and rules [3]. Decision tree is a data mining There are various ways of predicting solid method used for prediction and classification. waste generation that can be grouped into five While using these methods we can prevent key groups: descriptive models, analysis, the food from being wasted [3]. And the food flow of objects, the flow of time series, cooked is just the right amount [3]. methods of strategic planning strategies [1]. A research is carried by the students of the The vector support machine (SVM) and its science department of university „Degli Studi nearest neighbors to k (kNN) should use di udine‟ [5]. They provided statistics relating algorithms of machine sensors to test their to consumer attitude and behavior towards the predictive capacity for food waste produced food they consume. The techniques they [1]. These are the machine learning techniques adopted for the same is: 469  Aiming at understanding the issue of consumer food waste [4]. First of all the food wastage at household level. authors received relevant studies based on the  Proposing a hypothesis based upon purpose of review to reduce the evidence of people‟s attitudes and individual why food is being created in the area [4]. In behavior. the literature review the authors used information such as Web Science, Scopus and They researched the key explanatory Google Scholar and reviewed 60 articles factors when predicting the possibility of food [4][9]. being eaten in the family, emphasizing a In addition, the collected papers are coded model for this [5]. and the coded papers are compiled around the Techniques followed in the paper essentials that are considered and the items “Hospitality Restaurant Operations in Regard that seem to affect the amount of food to Food Security, Through Food Waste and delivered to the family centers and are Loss Control Mechanisms” [6]- assessed for equality.  Optimized quantities Rewritten documents are originally  Waste tracking and analytics converted into three domain types [4]: i. Social class features Portion Choices, Customized Dishes & ii. Psychological and social factors Smaller Plates iii. Behavior related to family food standards 2.1.1. How? 2.2. Problem Formulation To this end, they conducted a survey on the distribution of a questionnaire in homes [5]. The basic problem starts at a very basic level, Data obtained over a two-month period [5]. that is food wastage. According to some They used data collection methods [5]: studies [8], if we divide the total food  Online questionnaire forms produced by the planet earth into 3 equal parts  Face to face questionnaires then we are easily wasting one part of that. Food wastage has been such a common topic The number of responses they got are since the time we were born that around 500 which is a complete dataset with even now at such an advanced level, we clean and relevant values and without any null still don‟t have any solution to this problem. values [5]. The problem statement for our project is Work includes in paper “Hospitality mainly to manage the food wastage by making Restaurant Operations in Regard to Food new models in artificial intelligence to solve Security, Through Food Waste and Loss this problem. But for the very least, what we Control Mechanisms” [6]- have thought to acted upon is to implement a  By tracking the food wastage, we can model on our university‟s hostels. order the amount we need from What we plan to do is to solve the food suppliers [6] wastage in the hostels by making a new model using artificial intelligence.  As everyone's preferences are different so we can make customized dishes for them [6]. 2.3. Proposed Scheme  Provide small amounts of standard menu item with fill option [6]. The proposed system is to form a new model for food wastage management which The strong data available makes a case for mainly deals with the technological aspects in investing in food waste prevention efforts [6]. solving the issue of food wastage in the university/college hostel mess at individual 2.1.2. Methods? level. The proposed idea can be implemented by any university/college in their concern hostel messes in order to save food without This paper reviewed the playful but fast- being wasted. growing body of educational textbooks on 470 As an overall impact the system will consist of set of algorithms best suited for problems of these kind and a model by which many future problems could also be solved by the saving the food from being wasted. The model will be a general logical system of a type which can work with distinct datasets and problems. To create a model that could detect the patterns on how much a particular item is made for one-time meal so that we can save maximum amount of food and how much it is being wasted by the faculty and students. In this project we will first calculate the number of plates used by creating an Arduino based system, which will count the number of plates passing through it and this process will go on for about weeks including the number of students who eats in the mess. By knowing the number of absent student‟s waste food by not eating in mess we can decrease the quantity of food being prepared Simultaneously we will calculate the total amount of particular items being wasted and will calculate its weight. Once we get a sufficient amount of data to work upon, we will process our data and clean it. These are the steps to get the data on which we will work on them. After this, we will work Fig: User Flow Chart for Mess Management on the data and create a machine learning model that can give us the desired. We will use the mixture of algos which gives us the best 3. Research Objectives results. Once we analyse the result we will implement and observe how much food is The proposed research is aimed to decrease wasted if it is greater than the limit, say 9 kg, the wastage of food by the modern means of then we will again put the input in our model artificial intelligence. The proposed aim will and it will go on till we reach a less than a be achieved by dividing the work into limit. following objectives: 1) Data collection: The most important step in 2.4. Methodology all of the steps. It is to attain all the information related to food wastage in the The following methodology will be hostels. How we will do that is by followed to achieve the objectives defined for surveying and observing the food wastage proposed research work: every day when students and staff i. Detailed study of food wastage in the members are to have their lunch (we hostel mess will be done. choose this time because being present in ii. Data would be collected and then the mess in the morning or night for just cleaned to be used. the data collection will not be easy). Still iii. Various parameters like artificial we will try to collect all the data regarding intelligence will be identified to the topic according to the menu of every evaluate the proposed system. day. iv. Comparison of new implemented 2) Data integration and cleaning: Integrating approach with existing system all the data collected together and to clean approaches will be done. the data for making it trustable and 471 consistent. Without a good dataset, it will provisioning yet in addition in food just be pointless to work upon the problem. squander generation [4]. 3) The next step is bit of a technical step, The results are made on dataset that is where we will just choose an algorithm to collected from the journal “The quantity of work and interact with our dataset for food waste in the garbage stream of southern gaining insights and for gaining otherwise Ontario, Canada households” which contains negligible (but important) data. 86 rows and 18 columns. The dataset includes The final step would be to make a model all the household waste that is produced by the which can solve the problem we are facing and Canada households. This research paper what we really want to work upon is on mainly focuses on the food wastage that is making a model which can solve other being disposed in the garbage stream. problems with some constraints. The data was collected during year 2012- 2015 where it includes the household wastage 4. Results that is gathered from 9 municipalities of Southern Ontario which includes twenty-eight single-family households. Then the data was Derived from the papers that were aggregated and analysed as to expand the food researched as mentioned in the table for this waste estimates in the garbage stream research paper the results stated that most of considered for the study[11]. the hostel mess experienced comparative Every one of the 28 datasets comprise of waste examples [2]. Vegetarian main courses waste organization study information from 100 dips and were the most squandered [2]. families. Normally, each example of 100 Checked isolation decreased food wastage [2]. families is incorporated from 10 inspecting 33% of the expense of food is squandered [2]. zones of 10 continuous homes deliberately Planning and shopping routines are important chose by the particular region to work as an predictors of food wastage behaviour [5]. agent test. One region spoken to by five Socio-economic and segment family unit inspecting territories of 20 homes. Along these attributes are critical factors [5]. Factors like lines, there were a sum of 85 inspecting mentalities, age and pay influence squander regions over the nine regions. Every district conduct essentially [5]. Lack of attitudinal and chooses their diverse examining regions control beliefs [5]. It could be helpful to dependent on elements, for example, lodging expand collaboration between food esteem type (e.g., more established homes, more up to chain factors [5]. Describing food waste date homes) and neighbourhood financial behaviour and practices and perceptions [4] status. The examining zones are spread out  Psychology-situated methodologies give over week by week squander assortment days bits of knowledge into purchaser concerns, and commonly 2 to 4 testing regions are inspirations and standards around food gathered every week day. Squander tests are squander and their causal relationship on gathered from examining regions on their expectation to lessen food waste and (self- waste assortment day and are blocked at the detailed) conduct [4]. control before city assortment. The examples  Generally, consumers consider throwing are taken to an arranging territory and are away food as improper behaviour, the vast arranged into upwards of 120 arranging larger part of family units indicate that classifications, including a solitary "food they are at any rate to some degree squander" classification. The arranged food worried about discarding away food [4]. squander is gauged and recorded. Assortment  Implicating about food waste is a and arranging of squanders is attempted by significant predictor of food squander squander reviewers (i.e., organizations that reduction and plays an important job in the give proficient waste piece study expectation to reduce food waste [4]. administrations to regions). Each waste  Food-related household practices and creation study is rehashed twice more than two routines- the complex essence of food back to back a long time for similar families. waste, household routines such as Along these lines, two week after week planning, shopping, putting away, information focuses (i.e., week 1 and week 2) cooking, eating, and overseeing extras make up the normal of each testing territory's assume a conclusive job in food occasional information point. Squander 472 creation examines are rehashed up to 4 times The above figure is stacked area chart each year (i.e., to include every one of the four which is made by Winter Week 1(kg), seasons) for a similar testing regions and Winter Week 2(kg), Spring Week 1(kg), family units[10]. Spring Week 2(kg), Summer Week 1(kg), Moreover, we ordered information on a few Summer Week 2(kg), Fall Week 1(kg), factors which might impact the assessments of Fall Week 2(kg) columns in the dataset. food garbage removal for incorporation as free factors in factual models. For every one of the Where every zone of shading speaks to examples, we recorded the waste evaluator, one piece of the entirety. The parts are period of each examination (i.e., winter, stacked up, generally vertically. The spring, summer, fall), testing zone type (i.e., tallness of each shaded stack speaks to the urban, or country), and family unit access to rate extent of that class at a given point in food squander redirection programs (i.e., green time. A stacked zone diagram may be container program for gathering food utilized to show the breakdown of help for squanders at the control). Moreover, appraisals various ideological groups after some of the quantity of individuals per family and time. The stacked area chart displays how middle family salary (Canadian dollars) were the amount of food waste that is being aggregated for each example region utilizing produced of the households which are information from the 2011 Canadian registration at the scattering region level, considered for the survey. which is the littlest areal unit for which Statistics Canada discharges segment information and is a solid intermediary for each inspecting territory. By analysing the data and graphs, we discovered the patterns in our data which results in knowledgeable discovery. By finding out the solutions we can implement on a small scale at first and if it's successful then, we will try to make our model close to accurate based on the quality data which follows the V's of big data. Following are the data visualizations which are made in Jupyter notebook using Fig: Count Catplot Seaborn Graph various python libraries such as pandas, Count catplot plot basically plots the seaborn, plotly etc after pre-processing quantity of perceptions in each straight out techniques applied on our sample data so that factor with a bar, the above figure represents a patterns about the kind of data and the patterns caplot which shows the relationship between a of food waste can be noticed. numerical and one or more categorical variables using one of several visual representations on columns GB/nGB and Food waste sorted into Avoidable/Unavoidable Categories. The below figure represents the cluster points charts which is a clustering chart, where found the average waste engendered by each student with which we can analyse the minimum quantity of pabulum that can be preserved. Fig: Stacked Area chart 473 variable. Seaborn countplot is a barplot where the needy variable is the quantity of occurrences of each occasion of the free factor. The above figure represents a countplot which is kind of like a histogram or a reference chart for some clear-cut territory, here we considered GB/nGB and urban/rural columns. It basically shows the quantity of events of a thing dependent on a particular kind of classification on columns GB/nGB along the rural and urban columns in the dataset considering the household count. Fig: Cluster points M1 = (total-(food_wasted+food_left)) M1, M2, M3…………………. M= (M1+M2+………+Mn) / N Taking mean of food not used per day or months as per requirement. Applying the model on our data and with the avail of k-denotes we can engender clusters and find the center points which is most proximate point to all other values as shown in fig cluster points. We abbreviated the amplitude of pabulum being made on daily substratum and accumulated the data of aliment wasted or leftover. We applied the model on data and will go on until we reach the wastage of aliment at its min. Fig: Distplot Seaborn Graph A distplot plots a univariate distribution of observations. Seaborn distplot lets us show a histogram with a line on it. The above figure is Seaborn Distplot which allows us to show a histogram with a line on it. This can be shown in all kinds of variations. Here we considered Winter Week 1(kg), Winter Week 2(kg), Spring Week 1(kg), Spring Week 2(kg), Summer Week 1(kg), Summer Week 2(kg), Fall Week 1(kg), Fall Week 2(kg) columns in the dataset to show how the observations are made. Fig: Count plot Seaborn Graph 5. Conclusion The final solution is not only using Structured presentations are valuable for information alone by using various showing connections between all out technological aspects alone in order to create information and in any event one numerical 474 behavioural change to save the food from their policy implications. Journal of Cleaner being wasted, but is also majorly depends on Production, 182, 978-991. the challenging habits of the users and the [5] Marangon, F., Tempesta, T., Troiano, working of the mess management in the S., & Vecchiato, D. (2014). Food waste, universities/colleges. The current paper set out consumer attitudes and behaviour. A study in to audit observational, peer-checked on the North-Eastern part of Italy. Italian Review concentrates on family units' food squander of Agricultural Economics, 69(2-3), 201-209. rehearses, and distil socio-segment and [6] Okwachi, S. W., Miricho, M., & psycho-social factors just as food-related Maranga, V. (2018). A Study Focus in to family rehearses. Hospitality Restaurant Operations in While stressing the procedures that can be Regard to Food Security, Through Food embraced by people to forestall food squander Waste and Loss Control Mechanisms (No. in their family units, one should nonetheless, 458). EasyChair. recognize the person as inserted in more [7] Pinstrup-Andersen, P. (2000). The extensive social, economic, and social future world food situation and the role of structures that may forestall the reception of plant diseases. Canadian Journal of Plant less inefficient practices. Besides, insufficient Pathology, 22(4), 321-331. time to think about food all in all, and food [8] Depta, L. (2018). Global food waste squander specifically, combined with the and its environmental impact. apparent unpredictability of everyday lives [9] Makkar, A., & Kumar, N. (2020). An may transform food squander counteraction efficient deep learning-based scheme for web into an overwhelming undertaking. Yet, there spam detection in IoT environment. Future has been little examination led on how seen Generation Computer Systems. time accessibility influences individuals' waste [10] Makkar, A., Garg, S., Kumar, N., practices. On the off chance that we are to Hossain, M. S., Ghoneim, A., & Alrashoud, handle food squander in a deliberate manner, M. (2020). An Efficient Spam Detection we should likewise consider. Accordingly, an Technique for IoT Devices using Machine all-encompassing food squander counteraction Learning. IEEE Transactions on Industrial approach needs to go past putting the duty Informatics. exclusively on people. In the quest for [11] Makkar, A., Kumar, N., Zomaya, A. Y., arrangements, more mindful and competent & Dhiman, S. (2020). SPAMI: A Cognitive purchasers are required as much as submitted Spam Protector for Advertisement Malicious strategy producers who are eager to actualize Images. Information Sciences. the correct blend of strategy measures to make squander counteraction the favoured alternative for lodging wrecks just as family units. References [1] Abbasi, M., & El Hanandeh, A. (2016). Forecasting municipal solid waste generation using artificial intelligence modelling approaches. Waste management, 56, 13-22. [2] O'Shea, C. E., Merritt, K. A., & Cooney, R. J. (2017). Reducing Food Waste at IIT Mandi. [3] Bozkir, A. S., & Sezer, E. A. (2011). Predicting food demand in food courts by decision tree approaches. Procedia Computer Science, 3, 759-763. [4] Schanes, K., Dobernig, K., & Gözet, B. (2018). Food waste matters-A systematic review of household food waste practices and 475 Table: Research Papers SN Paper Name Authors Techniques Methods 1 Forecasting Maryam Descriptive statistical Support vector machine municipal solid waste Abbasi, Ali El models, relapse (SVM) and k-nearest generation using Hanandeh investigation, Regression neighbors (kNN) are two artificial intelligence strategy, time series intelligent Machine learning modelling examination and system algorithms. approaches Artificial-Intelligence techniques 2 Reducing Food Ryan They adopted simple Analyzing every day Waste at IIT Mandi Cooney, Harsh Techniques like present wastage according to the Gupta, and absent surveillance. number of meals taken. Kathryn, Merritt, Colleen O’Shea, Raghav Sethi 3 Predicting food Ahmet Data mining and Data mining is basically demand in food Selman decision trees are used collecting all the data and courts by decision Bozkir, Ebru analyzing the data which is in tree approaches Akcapinar huge amount in order to get Sezer the meaningful patterns and rules 4 Food waste Karin For the literature Initially reviewed codes matters - A Schanes, Karin search the authors used scaled up into three core systematic review of Dobernig, databases such as Web of types: household food Burcu Gozet Science, Scopus, and i. Socio-demographic waste practices and Google Scholar and factors their policy reviewed 60 articles. ii. Mental-socio factors implications iii. Food-according family-level behaviors 5 Food waste, Francesco Providing statistics Aiming at understanding the consumer attitudes Marangon, relating to customer issue of food wastage at and behaviour. A Tiziano demeanor and conduct household level. study in the North- Tempesta, towards the food they Proposing a hypothesis based Eastern part of Italy Stefania consume. upon people’s attitudes and Troiano, individual behavior Daniel Vecchiato 6 Hospitality Simon Optimized quantities, By tracking the food Restaurant Were waste tracking and wastage, we can order the Operations in Regard Okwachi, analytics, portion choice. amount we need from to Food Security, Moses suppliers. Through Food Waste Miricho and Providing smaller amounts of and Loss Control Vincent a common menu thing with Mechanisms Maranga the choice of tops off.