 Hello, go ahead sir go ahead you can go you can start yeah so hello dot dot.com 2018 2019 this is Shorup Choudhary from Bangladesh today my topic is over view on apriori machine learning actually apriori is a machine learning algorithm that create resolution rules from the item set so before going our main topic we to cover a small idea about what is artificial intelligence what is machine learning what is deep learning so what is artificial intelligence artificial intelligence is a technique which allow computers to mimic human intelligence including use of logic decision trace and machine learning including deep learning so what is machine learning machine learning is a part of an artificial intelligence which is a subset of artificial intelligence where statistical method are used to help a system improve a task with training and experience this category includes deep learning so what is deep learning deep learning is a subset of machine learning where a system can train itself to perform tasks including a language in machine analysis by using multi-layered neural networks so we see that deep learning is a part of machine learning and machine learning is also a part of artificial intelligence so here today our main topic is machine learning we are talking about apriori algorithm which is a machine learning algorithm so what is the apriori algorithm apriori algorithm is an unsupervised machine learning algorithm that generate resolution rules from the given data a solution rule implies that if an item air occurs then it can be also occurs with a certain probability most of the solution rules generated are in the if then format so here we see that there is a word which is unsupervised what is unsupervised unsupervised mean unlabeled data when any machine learning model learn from unlabeled data this is we called unsupervised machine learning unsupervised learning so going to our next slide for example if people buy a surface book then they also buy a surface book case to protect it for the algorithm to direct such conclusions it first observes the number of people who brought an iPad case while purchasing an surface book this way a ratio is derived out of the hundred people who purchased a surface book 85 people also purchased a surface book case next slide so there are some three key concepts of apriori algorithm which is frequent item set apriori property and joint operation what is frequent item set frequent item set is a set of item which is medium support what is apriori property apriori property is a subset of infrequent item set which must be frequent joint operation to find joint operation means when one or two items set create a new item set so this is we called joint operation suppose to find lk k is a set of candidate k item set is generated by joining lk minus one with itself applications of apriori algorithm so market basket analysis is one of the most common applications of apriori algorithm many e-commerce giants use apriori to draw data insights on which products are likely to be purchased together and which are most responsive to promotion for example a retailer might use apriori to predict people who buy sugar and flour are likely to in your picture we see that when we search e-commerce site for buying a Microsoft Surface Pro so we see that there are other similar product like that so in this picture we see that system show other similar Microsoft Surface Pro or Surface Book or something like that so this is called this is here actually apriori algorithm works here apriori algorithm create this item set and show the customer this product so another application of apriori algorithm is detecting adverse drug reactions apriori algorithm is used for association analysis on healthcare data like the drugs taken by patients characteristics of each patient adverse ill effects patient experience initial diagnostics etc these analysis produces association rules that help identify the combination of patient characteristics and medications that lead to adverse side effect of the drugs so actually here apriori algorithm create rules for the patients create rules from the drugs which is usable patients what is a side effect of these drugs how actually works for the patient of these drugs so here actually apriori algorithm work so now we are diving into how actually apriori algorithm work how it's work before studying that we need to know that what is input and what is output in a pre-algorithm in the approach algorithm input is a transaction database and a minimum support threshold set by the users and what is output output is a frequent data set so actually here input is transaction database and output is frequent and set so in our first step we see a table where there are four transaction and I can set our pasta lemon bread and orange so we see that each transaction have a different item set we see that in transaction one there are four I can set in transaction two there are two I can set three have three I can set and four have four I can set so in our item set we see that for pasta our support is four for lemon it is three for bread this one for orange it is three and for cake it is two but here our minimum support is two so in this step we need to find the item set which is below our minimum support so here we see that bread is our bread is a bread item set as minimum support is lower than our minimum support our minimum support is two and here bread is bread support is one so we need to cut down this item set and create a new table so here we see that for new item set we got pasta lemon orange and cake and each item set support value is equal to or greater than our minimum support in this step we see that each item set in this step a pre-algorithm will find a frequent item set containing two items to do that the pre-algorithm combines each frequent item set of size one each single item really to obtain a set of candidate item set of size two containing two items so here we see that from size one item set two we are creating such two items it from size two item set there we see that there are six item set in our next step we see that there are we need to find that is there any infrequent item set or not so we see that there is no infrequent item set so what is infrequent item set I told you you know in my previous slide so actually infrequent item set is the dough those item set which have a infrequent subset so there is no infrequent subset of this item set so there are no infrequent item set in my step five we see that we again need to we again need to calculate which item set support value is less than our minimum support so here we see that lemon and cake is lemon and cake because if support value is less than our minimum support which is one so we need to cut down this lemon and cake and create a new item set table so here we see that we got pasta lemon where support value is 3 pasta orange where support value is 3 also pasta cake is 4 is 2 sorry lemon orange is 2 and orange cake it is 2 so here we are creating support value we are taking those support value which medium which is equal to or greater than our minimum support in this step in step 8 in this step we need to find again is there any infrequent subset or not so here we see that now it is time we got two infrequent subset so which is pasta lemon cake and lemon orange cake because here lemon and cake are infrequent so we to again cut this item set and create a new item set table so from here we got two item set which is pasta lemon orange and pasta orange cake so in this step we again scan the database to calculate the exact support of the candidate item set of size 3 to check if they are really frequent so we see that our both item set support value is equal to our minimum support which is 2 so there is no problem in this step we again see that is there any infrequent subset or not so here we see that there are no infrequent subset because there are no item set which is infrequent so we are taking two of this item set so now in this three size 3 item set we are creating now size 4 item set which is pasta lemon orange and cake now we see that algorithm eliminates item set of size 4 having a size subset of size 3 because that is infrequent because we see that there is a lemon and cake so we know that a lemon and cake is infrequent subset so we need to cut down this item set so now we got that there are 11 item set this is our main final item set result so here we see that for pasta their support value is 4 for lemon it is 3 for orange 3 for cake is true pasta and lemon and pasta and orange both are 3 pasta and cake lemon orange orange cake pasta lemon orange and pasta orange cake they are also 2 so here we see that each item set support value is equal to or greater than our minimum support so there is no problem so this is our final item set so what is the advantages of apriori algorithm so actually it is implement is it is easy to implement and can be paralyzed easily apriori implementation make use of large item set properties so we see that when we are using apriori algorithm in e-commerce site or something like that we can easily implement it and it can it can be paralyzed easily rather than other machine learning algorithm then it is perfect for the as a rule-based machine learning model now we see a simple applications on the on based on this apriori algorithm so here we see that this is our applications this is our main algorithm so if we run our applications so we see that so we see that there is a four box which is one for transaction frequent item set another and another is frequent item sets and last one is confidence so here we see that there is a there is a option for our minimum confidence so here we select our medium confidence is 2 sorry here we select our medium confidence 60% and also we are selecting our minimum confidence 2 million support is 2 connection problem so here we are not our transaction data set so we see that our transaction data show in the transition box this pasta lemon plate and pasta lemon pasta orange cake and pasta lemon orange cake so now we see that if we change our minimum confidence into 61% and our minimum support we change our minimum support is 2 so now we are find our frequent items which is pasta lemon orange and cake and each item set value is a item set support below is equal to or greater than our minimum support which is see before in our slide and now we see that we check our confidence so this is our main decision rules for this item set direction so here we see that in confidence section from each item set confidence section is greater than equal to 60% so if we change our confidence value which is 46% so if we again check our confidence so here we see that there are many other more confidence association rules item set so which is pasta orange cake which is 50 minimum confidence 50% so this is our application so in our whole session we learned that what is actually a pretty good on how actually towards which kind of platform is perfect for a pretty rhythm we see that most of the cases are e-commerce sites for creating association rules which is perfect for this algorithm so so this is our main session hello sorry sir Rob we were trying to figure out the sound thing could you minimize that window for me please so we don't get the inception can everybody hear me okay out there yes everybody on the chat do you guys hear the echo anymore yep we still got echo all right sweet echo all the things go nice thank you for the sounding yes oh sounds great no echo no echo sweet thank you I thought there was an echo perfect thank you so much so Rob I don't think we got any questions so yeah I mean the epic when here I'm going to switch this over here to the Q&A so we both see each other the epic point of the day is that we don't have an echo and we can chat with you without deafening everybody so so Rob thank you so much now are you're you're a college student right now right yes right now I'm working as a Microsoft student partner awesome so so what are you doing as part of being a Microsoft student partner can you tell me a little bit about that we are conducting workshop about SEO or creating also a range workshop for Microsoft ImagineCup how a student use their implement their ideas by using SEO in ImagineCup that's awesome the one thing the thing I was going to say I loved was that you were using the Azure Labs to bring up your development and do everything there that was great so and the reason why I'm saying that is like you know guys we have a student here doing a presentation remotely all the way from India right you're doing all this for us and so we want you the community to share this kind of content with us right yeah we have the little hiccups with the sound and whatnot and people coming in now but you know what this is what makes the community fun right absolutely so thank you so much so Rob for taking the time to share your knowledge with us what we're going to do here is we're going to wrap up so we can kind of be caught up in time and so everybody out there watching the stream we're going to switch to our slate and we're going to get things going all right so so Rob thank you so much it was a great presentation and we'll be back here in about five minutes or so thanks everybody thanks