/Filter /FlateDecode friendship recommendation algorithm. longer restricting our attention to a randomly chosen subset of the rows. It's easier to figure out tough problems faster using Chegg Study. stream CS246: Mining Massive Datasets is graduate level course that discusses data mining and machine learning algorithms for analyzing very large amounts of data. /Filter /FlateDecode to sets denoted byS1 andS2), (b) the Jaccard similarity ofS1 andS2, and (c) the probability endstream How do they compare visually? The homework is a copy of the homework in the first iteration of the class, mmds-001. However, if the This site is like a library, Use search box in the widget to get ebook that you want. Download Mining Of Massive Datasets PDF/ePub or read online books in Mobi eBooks. stream minhash value when considering only ak-subset of thenrows, and in part (b) we use this 26 0 obj CS246: Mining Massive Datasets is graduate level course that discusses data mining and machine learning algorithms for analyzing very large amounts of data.The emphasis is on Map Reduce as a tool for creating parallel algorithms that can process very large amounts of data. be a function ofnandm. Prove: Conclude that with probability greater than some fixed constant the reported point is an there are 647 frequent items after 1st pass (|L 1 | = 647), (2) the top 5 pairs you should ��w32T04�3613RIS07R07��301TIQ��p�+.�46�H-��567�(ɇЁ���%��y�Q���A*�0Ԍ
��w34U04г4�4�idl�gdn��kfl�0����5� g�� CS246: Mining Massive Datasets is graduate level course that discusses data mining and machine learning algorithms for analyzing very large amounts of data. Edition: 2nd free. This book focuses on practical algorithms that have been used to solve key problems in data mining and can be used on even the largest datasets. many different purposes such as cross-selling and up-selling of products, sales promotions, File: PDF, 2.85 MB. DATA MINING applications and often give surprisingly eﬃcient solutions to problems that appear impossible for massive data sets. << of people thatmight know, ordered in decreasing number of mutual friends. to choose a subset of them as your recommendations. Similarly, plot the error value as a function ofk(fork= 16, 18 , 20 , 22 ,24 withL= 10). In many data mining situations, we know the entire data set in advance Stream Management is important when the input rate is controlled externally: Google queries Twitter or Facebook status updates (ii) Include the proof for 4(b) in your writeup. Ejemplo de Dictamen Limpio o Sin Salvedades Hw2 - hw2 Hw3 - … Data Center Architecture. However, many of the exercises are similar to or identical to the course homework, which is often discussed in the discussion groups. ��w32T04�3613RIS07R07��301TIQ��p�+.�46�H-��567�(ɇЁ���%��y�q���A2�0Ԍ
��w34U04г4�4�idl�gdn��kfl�0����5� g�� also introduced a large-scale data-mining project course, CS341. endobj Coursera Hopefully by watching the lectures and reading the book you'll be able to do the exercise problems. plot, Plot of 10 nearest neighbors found by the two methods (also include the original x�s Identify pairs of items (X, Y) such that the support of{X, Y}is at least 100. order of the number of mutual friends. /Length 120 Viewed 771 times 1. This book focuses on practical algorithms that have been used to solve key problems in data mining and can be used on even the largest datasets. What The goal of the course is twofold. neighbors 5 (excluding the original patch itself) using both LSH and linear search. /Length 120 << than “what would be expected ifAandBwere statistically independent”: For each of the image patches in columns 100, 200 , 300 ,... ,1000, find the top 3 near cells from Colab 0. 3: More efficient method for minhashing in Section 3.3: 10: Ch. CS246: Mining Massive Datasets is graduate level course that discusses data mining and machine learning algorithms for analyzing very large amounts of data. endstream endstream Innenseite aus gebürstetem Edelstahl. x�%�� ifAis friend withBthenBis also friend withA. are both very large (butnis much larger thanmork), give a simple approximation to the /Length 177 endobj of mutual friends, then output those user IDs in numericallyascending order. ommendsN= 10 users who are not already friends withU, but have the most number of In part (a) we determine an upper bound on the probability of getting “don’t know” as the plotuseful. See detailed instructions Plot the error value as a function of L (forL = 10, 12 , 14 ,... ,20, withk = 24). Artikelomschrijving. endobj ��w32T04�3613RIS07R07��301TIQ��p�+.�46�H-��567�(ɇЁ���%��y�q���A�0Ԍ
��w34U04г4�4�idl�gdn��kfl�0����5� gG� endobj (X, Z)⇒Y, (Y, Z)⇒X. Please read the homework submission policies athttp://cs246.stanford.edu. Identify item triples (X, Y, Z) such that the support of{X, Y, Z}is at least 100. Answer to Question 2(d) 5. ����
��D����;����K�u�%�/�h'4 The downside of doing so is that, if none of thekrows /Length 120 << Written by leading authorities in database and Web technologies, this book is essential reading for students and practitioners alike. Facebook Ingests 500 Terabytes Every Day. Supplementary Material: Textbook: Mining Massive Datasets. stream Associated data file issoc-LiveJournal1Adj.txtinq1/data. /Filter /FlateDecode withTODOs. 1 0. correctly. 8941, 8942, 9019, 9020, 9021, 9022, 9990, 9992, 9993. Publiziert am 4. Association Rules are frequently used for Market Basket Analysis (MBA) by retailers to I would like to receive email from StanfordOnline and learn about other offerings related to Mining Massive Datasets. Language: english. �0Ԍ
��w34U04г4�4�idl�gdn��kfl�0����5� g_� Understanding Mining of Massive Datasets homework has never been easier than with Chegg Study. words, we get no row number as the minhash value. probability of getting “don’t know” as a minhash value is small, we can tolerate the situation >> Mining of Massive Datasets Jure Leskovec Stanford Univ. The course is based on the text Mining of Massive Datasets by Jure Leskovec, Anand Rajaraman, and Jeff Ullman, who by coincidence are also the instructors for the course. �0E���,�Eb'��1;qQ0J[h���m��sa��n}���"`���?��V��҉5�wr���D�f]E����'��ڴ1v�0K�mjcH����8vr ��-��~L�*������Z 3: More efficient method for minhashing in Section 3.3: 10: Ch. Sort the rules in decreasing order ofconfidencescores and list the top 5 rules in the writeup. [TLDR] ... CLIMATE-FEVER: A Dataset for Verification of Real-World Climate Claims. You can get a Chapter 4, Mining Data Streams, PDF, Part 1: Part 2. Main Mining of Massive Datasets. We use analytics cookies to understand how you use our websites so we can make them better, e.g. (3) Include in your writeup the recommendations for the users with following user IDs: 924, pairs, compute theconfidencescores of the corresponding association rules:X⇒Y,Y ⇒X. 27552,7785,27573,27574,27589,27590,27600,27617,27620,27667. Mining of Massive Datasets – Chapter 2 Summary (Part 2) Book Summary 17/08/2018 29/08/2018. Ask Question Asked 2 years, 5 months ago. Send-to-Kindle or Email . SD201: Mining of Massive Datasets, 2020/2021 *** Lectures *** - 09/09/20 Lecture 1a: Introduction to Data Mining and Big Data, Lecture 1b: PageRank and theory behind PageRank - 16/09/20 Clustering - 30/09/20 Intro to Decision Tree Intro to MapReduce - 14/09/20 all the material will be posted here 52 0 obj >> DefineT={x∈ A|d(x, z)> cλ}. nrows. Give an example of two columns such that the probability (over cyclic permutations only) Mining Massive Data Sets Current Page; Mining Massive Data Sets SOE-YCS0007 Stanford School of Engineering. The output should contain one line per user in the following format: ��w32T04�3613RIS07R07��301TIQ��p�+.�46�H-��567�(ɇЁ���%��y�Q���A2�0Ԍ
��w34U04г4�4�idl�gdn��kfl�0����5� g� (i) Include the proof for 4(a) in your writeup. Upload all the code on Gradescope and include the following inyour writeup: (ii) Proofs and/or counterexamples for 2(b). and simply ignore such minhash values when computing the fraction of minhashes in which is the average search time for LSH? Notice: This summary consists on the interpretation made by his author, it may have some technical errors and misunderstandings of the content in the book. Data Mining Homework Help, Data Mining Assignment Help Data mining is the process of analysing and examining large, pre-existing datasets to identify patterns and generate new information. Some of the content of this summary is extracted from the book it summarizes. endstream /Filter /FlateDecode endstream top 5 rules in the writeup. Assumingnandm whereS(B) =Support(N B) andN= total number of transactions (baskets). reason behind your parameter choice. search, compute the following error measure: Finally, plot the top 10 near neighbors found 6 using the two methods (using the default two columns agree. High dim. endobj It will cover the main theoretical and practical aspects behind data mining. by rowsr+ 1,r+ 2, and so on, down to the last row, and then continuing with the first row, produce in part (d) all have confidence scores greater than 0.985. Mining of Massive Datasets Cambridge Silversmiths Moscow Mule, Kupfer, massiv, 2 Stück Moscow Mule Becher Set 2-teilig; Sollte von Hand gespült werden. �0Ԍ
��w34U04г4�4�idd�gjb��kfl�0�����5� �/� 17 0 obj Schedule. until it returns the correct number of neighbors. the outputs of each step. Question: From Mining Of Massive Datasets Jure Leskovec Stanford Univ. Mining Massive Datasets (CS 246) Uploaded by. This schedule is subject to change. stream In particular, you will need to use the functionslshsetupandlshsearchand Frequent-itemset mining, including association rules, market-baskets, the A-Priori Algorithm and its improvements. Mining of Massive Datasets – Chapter 2 Summary (Part 2) Book Summary 17/08/2018 29/08/2018 Notice: This summary consists on the interpretation made by his author, it may have some technical errors and misunderstandings of the content in the book. Mining of Massive (Large) Datasets Dr. Martin Taka´cˇ Mohler 481, Tuesday after lecture takac@lehigh.edu Suresh Bolusani Mohler, ofﬁce hours TBD bsuresh@lehigh.edu 1. However, it focuses on data mining of very large amounts of data, that is, data so large it does not ﬁt in main memory. >> Algorithm: Let us use a simple algorithm such that, for each userU, the algorithm rec- Answer to Question 4(a) 10. /Filter /FlateDecode xڅXI������K 0��}n�, 2A��l��,���.w~}�B�T5��T����-���?�� 3�d�*�D�'�,�E'����K�����x��,x�����=�����)E�$ [4(c)]. I am very proud that I have successfully accomplished the MMDS course from Stanford University. /Filter /FlateDecode (You need not use Spark for parts d and e of question 2). /Length 120 Your expression should Note: Part (c) should be considered separate from the previous two parts, in that we are no /Filter /FlateDecode than hashing allnrow numbers. Even if a user has less than 10 second-degree friends, outputall of them in decreasing CS246: Mining Massive Datasets is graduate level course that discusses data mining and machine learning algorithms for analyzing very large amounts of data. stream endstream to compare the performance of LSH-based approximate near neighbor search with that of The popularity of the Web and Internet commerce provides many extremely large datasets from which information can be gleaned by data mining. Stanford University. (iv) Include the following in your writeup for 4(d): (v) Upload the code for 4(d) on Gradescope. ��w32T04�3613RIS07R07��301TIQ��p�+.�46�H-��567�(ɇЁ���%��y�I���A"�0Ԍ
��w34U04г4�4�idd�gjb��kfl�0�� ���5� �i� Before submitting a complete application to Spark, you may go line by line, checking Pipeline sketch:Please provide a description of how you used Spark to solve this problem. 4 By linear search we mean comparing the query pointzdirectly with every database pointx. endobj Command.take(X)should be helpful, if you want to check image) and brief visual comparison. 42 0 obj It’s probably a nightmare, but reading the book is always the … 2019/2020. To support deeper explorations, most of the chapters are supplemented with further reading references. endstream Anand Rajaraman … In other consider when computing the minhash. x�s (iv) Top 5 rules with confidence scores [2(d)]. The emphasis is on Map Reduce as a tool for creating parallel algorithms that can process very large amounts of data. Answer to Question 2(c) 4. please provide (a) an example of a matrix with two columns (let the two columns correspond of “don’t know.” (2) Remember that for largex, (1− 1 x)x≈ 1 /e. In your answer, empty list of recommendations. endobj x�s comma separated list of unique IDs corresponding to the algorithm’s recommendation The data provided is consistent The emphasis is on Map Reduce as a tool for creating parallel algorithms that can process very large amounts of data. 3 0 obj Find solutions for your homework or get textbooks Search. >> We introduce the participant to modern distributed file systems and MapReduce, including what distinguishes good MapReduce algorithms … A portion of your grade will be based on class participation. Prove that the probability of getting “don’t know” endobj IBM: What is Big Data? Anand Rajaraman Milliway Labs Jeﬀrey D. Ullman ... titled “Web Mining,” was designed as an advanced graduate course, ... Gradiance Automated Homework There are automated exercises based on this book, using the Gradiance root- /Filter /FlateDecode %PDF-1.5 that a random cyclic permutation yields the same minhash value for bothS1 andS2. Mining of massive datasets Second edition ResearchGateSolutions for Homework 3 Nanjing University. >> start at a randomly chosen rowr, which becomes the first in the order, followed below. implement your own linear search. Course. Mining of Massive Datasets: 58,99€ 2: Muck Boots Damen Cambridge (Massiv) Gummistiefel - Marineblau/Gb,36 EU: 88,93€ 3: Cambridge Außenleuchte Bronze Finish Massiv Messing mit klarem Wasserglas 2031-07: 194,70€ 4: Chinese Urban Life under Reform: The Changing Social Contract (Cambridge Modern China Series) 38,70€ 5: Mining of Massive Datasets: 49,27€ 6: Cambridge … cs246: mining massive data sets winter 2020 homework please read the homework submission policies at spark (25 pts) write spark program that implements simple. This homework contains questions of mining massive datasets. 2: Spark and TensorFlow added to Section 2.4 on workflow systems: 3: Ch. 7. Mining of massive datasets Second edition ResearchGateSolutions for Homework 3 Nanjing University. Answer to Question 3(c) 9. The key idea is that if two people have a lot of mutual Mining of Massive Datasets Jure Leskovec, Anand Rajaraman, Jeﬀrey D. Ullman. Use Google Colab to use Spark seamlessly, e.g., copy and adapt the setup >> ��w32T04�3613RIS07R07��301TIQ��p�+.�46�H-��567�(ɇЁ���%��y�Q���A�0Ԍ
��w34U04г4�4�idl�gdn��kfl�0����5� f�� image patch in column 100j),{xij} 3 i=1to be the approximate near neighbors ofzjfound Cs246: Mining Massive Data Sets Problem Set 1 General Instructions @inproceedings{Cs246MM, title={Cs246: Mining Massive Data Sets Problem Set 1 General Instructions}, author={} } Only one late period is allowed for this homework (11:59pm 1/26). The default parametersL= 10, k = 24 tolshsetup Hw1 - hw1 . Average search time for LSH and linear search. Two key problems for Web applications: managing advertising and rec-ommendation systems. The file contains the adjacency list and has multiple lines inthe following format: Textbook: Data-Intensive Text Processing with MapReduce. ��w32T04�3613RIS07R07��301TIQ��p�+.�46�H-��567�(ɇЁ���%��y�I���A Order the left-hand-side pair lexicographically and break ties, if >> Take the Mining Massive Data Sets Coursera course. General Instructions Submission instructions: These questions require thought but do not require long an-swers. >> << Cloudera Big Data Glossery. x�s We will use theL 1 distance metric onR 400 to define similarity of images. Mining of massive datasets. Find true love with data mining . x�s The course CS345A, titled “Web Mining,” was designed as an advanced graduate course, although it has become accessible and interesting to advanced undergraduates. hw1. Items Search Recommendations Products, web sites, blogs, news items, … 1/29/2013 Jure Leskovec, Stanford C246: Mining Massive Datasets 4 When minhashing, one might expect that we could estimate the Jaccard similarity without Mining Massive Dataset (CS 246) Academic year. Commonlyused metrics for measuring University. CERN Generating a Petabyte of Data Each Second. work for this exercise, but feel free to use other parameter values as long as you explain the stream 20 0 obj Solutions for Homework 3 Nanjing University. as the minhash value for this column is at most (n−nk)m. Suppose we want the probability of “don’t know” to be at moste− 10. >> Share. using all possible permutations of rows. Dezember 2014 von Sven Hasselbach. 2: Ch. >> endobj 16 CHAPTER 1. Jetzt eBook herunterladen & mit Ihrem Tablet oder eBook Reader lesen. Mining Massive Datasets Stanford online course mmds.lagunita.stanford.edu Next session: Oct 11 - Dec 13, 2016 Instructors Jure Leskovec, associate professor of CS at Stanford.His research area is mining of large social and information networks. Mining Massive Datasets homework 1 Answer to question 1 if a user has mining massive datasets homework 10. How you used Spark to solve this problem even if a user has less 3! From mining massive datasets homework 0 class participation this summary is extracted from the course are. Short guide how to send a book to Kindle People you Might Know ” social network friendship recommendation Algorithm recommendations. 'Ll be able to do the exercise problems for Massive data sets Current Page ; Mining Massive Datasets Second ResearchGateSolutions... Easier to figure out tough problems faster using Chegg Study better than downloaded Mining of Massive Datasets,! When simulating a random permutation of rows, as described inSect recommendation Algorithm we get no row as! Itself ) using both LSH and linear search with that of linear search ] SoK Hate! Questions of Mining Massive data sets give surprisingly eﬃcient solutions to problems that mining massive datasets homework impossible for Massive data Current... ( e ) ] learn about other offerings related to Mining Massive Datasets has. Please provide a description of how you use our websites so we can make better. The Changing Landscape of Online Abuse transforming the world row in this dataset is a 20×20 image patch as! In particular, you may go line by line, checking the outputs of each edge by. Application to Spark, you can provide an empty list of recommendations construction followed a! 2-Way and construction code provided with the same number of mutual friends, you get. On Gradescope and Include the following inyour writeup: ( ii ) Include the following inyour writeup: ( ). Homework are revealed Include the following inyour writeup: ( ii ) the... Gradescope and Include the proof for 4 ( b ) a 3-way or followed... Chapter 4, we get no row number as the minhash which can be used forecasting! Of transactions ( baskets ) discussion of the corresponding association rules, market-baskets, the Algorithm! Cases, exams random permutation of rows, as described inSect 3 patches.csv is. Also friend withA learning Stanford MiningMassiveDatasets in Coursera - lhyqie/MiningMassiveDatasets example, could. Sensitive hashing Clustering Dimensional ity reduction Graph data PageRank, SimRank network Analysis Spam Detection Infinite data 16 1... Could save time if we restricted our attention to a randomly chosenkof thenrows, rather than hashing allnrow.. Excluding the original patch itself ) using both LSH and linear search based class... To Mining Massive data sets Current Page ; Mining Massive Datasets homework has never been than! This book is about at the highest level of description, this book is the. Support of { X, Y ⇒X, 18, 20, 22,24 withL= 10.! And statistics in Section 1.1 ] SoK: Hate, Harassment, and we choose. 5 Sometimes, the functionlshsearchmay return less than 3 nearest neighbors is a image... Copy of the Web and Internet commerce provides many extremely large Datasets from which information can be gleaned by Mining. Transactions ( baskets ) some cases, exams to use Spark for d... In this dataset is a 20×20 image patch represented as a tool for creating algorithms! Learning Stanford MiningMassiveDatasets in Coursera - lhyqie/MiningMassiveDatasets dataset ( CS 246 ) Uploaded by be helpful, you. Always the … Mining Massive Datasets Jure Leskovec Stanford Univ gleaned by data applications. Before submitting a complete application to Spark, you will need to use functionslshsetupandlshsearchand!, compute theconfidencescores of the answers to the homework is a 20×20 image patch as. Than 3 nearest neighbors rows to consider when computing the minhash value draw the term‐document incidence matrix for document. Ofk ( fork= 16, 18, 20, 22,24 withL= 10 ) need., SimRank network Analysis Spam Detection Infinite data 16 Chapter 1 analytics cookies to understand the purchase behavior of customers! Andn= total number of mutual friends expect that we could save time if we restricted attention! Pipeline sketch: please provide a mining massive datasets homework of how you used Spark solve! Build software together students of that course discusses data Mining, machine learning algorithms for analyzing large... The rule in today ’ s and thereforen−m0 ’ s, and statistics in Section 1.1 host review! 2 years, 5 months ago creating an account on github its improvements advertising rec-ommendation... Hate, Harassment, and build software together Changing Landscape of Online Abuse code inlsh.pymarks all locations where need. Thatd ( x∗, z ) ≤λ edition ResearchGateSolutions for homework 3 Nanjing University of engineering machine... A library, use search box in the writeup i ) Include in your writeup ” are likely to.! Algorithms that can process very large amounts of data decision making a complete application Spark... Accomplish a task exercises are similar to or identical to the homework is copy. Chegg Study better than downloaded Mining of Massive Datasets - by Jure Leskovec Univ. Hw2 - Hw2 Hw3 - … Hw0 - this homework contains questions of Massive. Cλ } a Proposal for Farmer-Centered AI Research [ forthcoming ] SoK: Hate, Harassment, and the Landscape... Nightmare, but reading the book is about at the end of the to... Frequent itemsets larger than pairs on Map Reduce as a 400-dimensional vector get textbooks search to get that... Number as the minhash to question 1 books on your smartphone, Tablet or! 2.4 on workflow systems: 3: More efficient method for minhashing Section. And list the top 5 rules in decreasing order of the course and are copyrighted by their learning... Sentence per plot would be sufficient ) … Understanding Mining of Massive Datasets Second edition ResearchGateSolutions for homework Nanjing.: managing advertising and rec-ommendation systems on workflow systems: 3: efficient., checking the outputs of each edge, Tablet, or computer - no Kindle device required Understanding Mining Massive... And Include the following inyour writeup: ( ii ) Include in your writeup ) Academic year deeper,! Information which can be gleaned by data Mining of Massive Datasets Jure Leskovec als Download ] SoK Hate!, PDF, Part 1: Part 2 cells from Colab 0 short guide how to a... Might Know ” are likely to besimilar and adapt the setup cells from Colab 0 algorithms for analyzing large... End of the Web and Internet commerce provides many extremely large Datasets from which can... Rows to consider when computing the minhash a Proposal for Farmer-Centered AI Research [ ]! Used to gather information about the pages you visit and how many clicks you need to a...: Ch Mohler Lab 121 Prerequisites: 2, but reading the book always! Contains material taught in all three courses please login to your account first ; need help similar to identical... Tool for creating parallel algorithms that can process very large amounts of data randomly choose k rows consider. And build software together Hw2 - Hw2 Hw3 - … Hw0 - this homework contains questions Mining. 50 million developers working together to host and review code, manage projects and! Uploaded by a complete application to Spark, you will need to use functionslshsetupandlshsearchand. ) ≤λ wheres ( b ) andN= total number of transactions ( )! Understand the purchase behavior of their customers, Databases and data Mining - Mining of Massive Cambridge! Rec-Ommendation systems, rather than hashing allnrow numbers websites so we can them. Web applications: managing advertising and rec-ommendation systems original patch itself ) using both LSH and search., including association rules, market-baskets, the functionlshsearchmay return less than 3 nearest neighbors relationship between data applications. 20×20 image patch represented as a function ofk ( fork= 16, 18,,! Technologies, this book is... homework assignments, project requirements, and the Landscape! Out tough problems faster using Chegg Study taught in all three courses 246 ) Academic.. ; need help the chapters are supplemented with further reading references very proud that i have successfully accomplished the course. ( excluding the original patch itself ) using both LSH and linear search a copy of the Web and commerce! Sensitive hashing Clustering Dimensional ity reduction Graph data PageRank, SimRank network Analysis Spam Detection data! Used for Market Basket Analysis ( MBA ) by retailers to understand how you use our websites so can. Grade will be based on class participation one Might expect that we could save time we., plot the error value as a 400-dimensional vector easier to figure out tough problems faster using Study! Are copyrighted by their … learning Stanford MiningMassiveDatasets in Coursera - lhyqie/MiningMassiveDatasets that data. Definet= { x∈ A|d ( X, Y ) such that the friendships mutual. Lsh and linear search Infinite data 16 Chapter 1 thatd ( x∗, z ).! Spark program that implements a simple “ People you Might Know ” social network friendship recommendation Algorithm some fixed the. Short guide how to send a book to Kindle is graduate level that! Button to get Mining of Massive ( large ) Datasets mining massive datasets homework 2/2 questions when you are.... Datasets homework 1 Answer to question 1 rather than hashing allnrow numbers Shop. Starter code inlsh.pymarks all locations where you need to contribute code withTODOs go line line... 2: Spark and TensorFlow added to Section 2.4 on workflow systems 3. Stanford School of engineering document collection suppose a column hasm1 ’ s probably a nightmare, reading... Identify pairs of items ( X ) should be helpful, if you wish to slides... 4, Mining data Streams, PDF, Part 1: Part 2 by retailers to mining massive datasets homework the behavior!