Business Case: Recommendation Systems Powered by AI—Still Room for Improvement
Much has been written about the wonders of some of the most well-known recommendation systems in use today at companies like Amazon, Netflix, LinkedIn, Facebook, and YouTube. These recommendations are credited with giving their companies a significant competitive advantage and are said to be responsible for significant increases in whatever system the company uses to keep score. For Amazon, that would be sales dollars. The Amazon recommendation system is said to be responsible for 35% of sales, a figure that has been cited by several authors dating back to at least 2013 (MacKenzie, Meyer, & Noble, 2013; Morgan, 2018). The Netflix recommendation system is also believed to be one of the best in the business. Netflix counts success in terms of how many shows people watch, how much time they spend watching Netflix, and other metrics associated with engagement and time on channel. But the Netflix recommendation system is also credited with moving dollars to the company’s bottom line to the tune of $1 billion a year (Arora, 2016).

In the realm of social media, score is kept a little differently, and in the case of Facebook and LinkedIn, recommendation systems are frequently used to suggest connections you might wish to add to your network. Facebook periodically show you friends of friends that you might be interested in “friending,” while on LinkedIn, you are frequently shown the profiles of individuals that might make great professional connections. Finally, YouTube’s recommendation system lines up a queue of videos that stand ready to fill your viewing screen once your current video finishes playing. Sometimes the relationship between your current video and the line-up of recommended videos is obvious. While watching a clip of a Saturday Night Live sketch, you can see that several of the recommended videos waiting for you are also SNL clips. But not always, and that is probably where some cool recommendation engine juju comes into play, trying to figure out what will really grab your interest and keep you on-site for a few more minutes, watching new clips and the increasingly annoying advertisements that now seem to find multiple ways of popping up and interrupting your use of YouTube’s platform without paying the price of admission.

While all of these companies are to be credited for pioneering recommendation technology that most likely generates beneficial results, it seems that more often than not, the recommendations we get are not as impressive as what so many blog writers would have us believe.

Today, all these recommendation systems have been infused and super-charged from their original creations with the power of artificial intelligence.

Answer the following questions:

• Has this really changed much in terms of the user experience?
• How many times do you really send a friend request to that person Facebook tells you that you share four friends in common?
• Would you accept a friend request from that individual if they sent one to you?
• How often do you try to connect with the professionals that LinkedIn recommends to you?
• Or do you find the whole process of deleting all those suggestions a pain?
• Finally, how often have you sat down to watch Netflix, and after scrolling through all their movies and television shows, you end up watching another channel or maybe decide to go read a book?
• Or when was the last time you purchased an unsolicited product that was recommended to you on Amazon?

Your paper should be 3 to 4 pages long using APA format. Provide appropriate citations

Do you need a similar assignment done for you from scratch? Order now!
Use Discount Code "Newclient" for a 15% Discount!

For the units 6-7 assignment you will be required to build and train a neural network to recognize letters of the alphabet and numbers based upon their design using a seven segment display where each segment of the display is one input into the neural network.

Using the numbering for the segments in the previous figure as the inputs into your neural network.  You are welcome to use any network design that will accurately solve the problem.   In this case your network must be able to identify the letter or number based upon the pattern of segments.   For example if the segments 1,2,3,4, and 7 are lighted then the pattern would represent the number 3.  The number three must be represented as a binary number that is the output of the neural network.

Not all of the letters of the alphabet can be accurately recognized.  The following chart represents the numbers and letters that your network must be able to recognize with the exception of letters S and Z which cannot be distinguished from the numbers 2 and 5.

The output of your network will be the ASCII code of the letter or number defined in the following list and represented as binary.  Since the largest number is 72 you will require 7 outputs to represent any of the 7 segment numerals.

Character       ASCII       Neural Network Output (Binary)

0                  48                0110000

1                  49                0110001

2                  50                0110010

3                  51                0110011

4                  52                0110100

5                  53                0110101

6                  54                0110110

7                  55                0110111

8                  56                0111000

9                  57                0111001

A                 65                1000001

B                 66                1000010

C                 67                1000011

D                 68                1000100

E                 69                1000101

F                 70                1000110

H                 72                1001000

We are using a 7 segment display to recognize letters and numbers, however it is important to know that we could use the same approach to recognize virtually anything.   We are using a 7 segment display because our simulated neural network only has the capacity to accommodate 10 input values.  However if we had more input values we could recognize any handwritten letter or number.  Consider the following examples example.  Assume that we had a matrix that was 16 x 16 (total of 256 point) and each of the points were an input into our neural network.  Further assume that we could overlay a hand written letter or number onto this matrix.  Every pixel that was colored by the letter would have a value of 1 and any pixel that was not colored would have a value of 0.

Using this as input we could easily train a neural network to ‘recognize’ various hand written letters.  As the amount of training increases the accuracy of the network to detect similar letters that might not have the exact same shape (but a similar one) would increase.

Let’s consider another example.   Assume that we had a neural network and instead of the inputs being 1 or 0, we had the ability to determine a value between 0 and 1 as input where 0 was the color white and 1 the color black and shades of grey as values in between 0 and 1.

Using this approach we could train the neural network to recognize faces, people, and objects … such as Albert Einstein here.  In practice, facial recognition software will typically limit the amount of ‘features’ that are used as input to the neural network to areas that have the greatest predictive power such as the position of the eyes, nose, mouth, and perhaps chin, but the principle remains the same as the network that we will build to recognize the letters in the 7 segment display.

To complete this assignment, you will need to download and install the Basic Prop neural network simulator.  Basic Prop is distributed as an executable Java Jar file.  You can either execute the file on your local computer or you can access the simulator in the Virtual Computing Lab.   To run the simulator on your local computer you will need to have the Java JRE (java runtime environment) version 1.5 or greater installed on your computer.

The simulator can be downloaded from the basic prop website at the following URL: http://basicprop.wordpress.com/2011/12/21/introducing-basic-prop-a-simple-neural-networ/

As part of the assignment you will need to create a pattern file for the 7 segment display input data.  You can see an example of a pattern file on the basic prop website.

I have provided an example of a pattern file below.  In this case I have set up the pattern file to teach the neural network to add two binary numbers.  Notice that the first set of digits is the input value in binary.  The second set of digits is the output value of the network.  Notice how in the first set of digits I have 0 0 0 0 1 0 0 0 0 0 1 as input.  In this example which adds two numbers together, I have defined the input as two binary numbers each 5 digits in length (in the first example the two numbers are 1 in binary).  The second set of digits are the output of the network in the example it would be 2 in binary (0 0 0 1 0) ….1 plus 1 is 2.

The network is then ‘taught’ to add the two numbers together.

Number of patterns = 3

Number of inputs = 10

Number of outputs = 5

[Patterns]

0 0 0 0 1 0 0 0 0 1    0 0 0 1 0

0 0 0 1 0 0 0 0 0 1    0 0 0 1 1

0 0 0 1 1 0 0 0 1 0    0 0 1 0 1

Another example is using the neural network to process logic patterns such as AND, OR, and the XOR. You can download the pattern file for AND logic and use the same file format for a network that can solve the 7 segment display problem (you could also cut and paste my add numbers pattern file from above).   When you have created the pattern file (the pattern file defines the number of inputs and the outputs in the problem and the number of test cases you can use it to train and test your neural network.

Using the simulator you will need to design a network, load the pattern file, and then train the network.  You should experiment with your network design to make sure that your network and solve all of the test cases.  If your network is not solving all of the cases, then you should alter the design of the network and train the network again.When you have workable network design that solves all of the cases:

• You should save your network, capture a screen shot of your simulator with the completed solution, and save your pattern file.
• You must report on the results of testing your network after it has been trained.
• You must include the average per pattern error metric as part of your testing results.
• Your error should be under 5% in order for the network to be acceptable.
• You must report the number of training steps that your network required.
• You should report the metrics chosen in basic prop for training your network report the defaults if you do not change them.
• You should report these metrics for EACH network that you test identifying the network that you selected as the solution for your assignment.
• You should also use the feature in basic prop to save your weights to a file.
• All of these items should be submitted as part of your assignment.
• You should also write a short paper explaining the process that you went through to develop your network including any network designs that were unable to accurately determine what each input character was and the reasons that you discovered the network design was not successful.

NOTE: You will have 2 weeks to compete this assignment; it will be due at the end of Week/Unit 7.

Do you need a similar assignment done for you from scratch? Order now!
Use Discount Code "Newclient" for a 15% Discount!

For many of us old school gamers, we started out playing good old fashioned board games. Many of us learned strategy from games like Risk and Battleship. For your project, you will be developing a simple battleship type game to test your knowledge of the concepts we have learned in Object Oriented Programming.

The purpose of this project is to have you design of a fairly complicated project using concepts we have learned and then implement the solution using some of the code that we wrote in the previous projects along with new code, and then test your game.

You need to first start with the design of the project. The design documentation, due as shown on the syllabus, will be worth one homework grade and when corrected and submitted will be 20% of your final project grade. If your design is carefully thought through, the coding of the game should be relatively straightforward since many of the classes and concepts closely match previous assignments. You will need to first read these requirements and make a design document (ensuring that all the requirements are met in the design). A sample design document is posted in Canvas and should be used as your design TEMPLATE (example). Create a design document complete with the class diagram and activity (flow) diagram, as well as any decisions you made on the best use of classes, inheritance, polymorphism, and exception handling.

After you have completed your design, then you will be ready to implement the game and test. I cannot stress enough that a good design document and understanding of the requirements will make the actual coding of the game much faster and more simple.

Please don’t leave things until the last two weeks. Get started now, and please ask your instructor for help BEFORE you get too lost. Get the big picture done first. Worry about the structure and implementation of the major functionality. Then if you have time, work on the little details, and minor error checking.

So now onto the requirements. You sunk my battleship!….

For your CSCI 2312 Project, you will develop a simple battleship game. Battleship is a guessing game for two players. It is played on four grids. Two grids (one for each player) are used to mark each players’ fleets of ships (including battleships). The locations of the fleet (these first two grids) are concealed from the other player so that they do not know the locations of the opponent’s ships. Players alternate turns by ‘firing torpedoes’ at the other player’s ships. The objective of the game is to destroy the opposing player’s entire fleet. In our game, ‘firing a torpedo’ will be allowing the player to take a guess at where on the grid their opponent may have placed a ship.

In the requirements, we will set forth other simplifying rules to limit the scope of this project.

Requirements

Given the requirements as a rough specification, you are to design the classes and implement the game. In our imaginary game company, the requirements below were developed by the Product Development Team and your instructor is the Product Owner. You are in full control of the choice of classes (please use classes appropriately or points will be deducted), data structures, algorithms, internal file format, detailed user interface scheme, or any other pertinent design decisions you need to make. As the Product owner, I care that it compiles and runs like it is supposed to, meets all the functionality and requirements I have set forth, and is easy to play and understand.

The Battleship game you are designing and implementing is a simplified version of the electronic Battleship game played in one player mode.

The game is played on four grids, two for each player. The grids are typically square and in our case will be 10 by 10. The individual squares in the grid are identified by the x coordinate (indicated by a letter) followed by the y coordinate (indicated by a number). The following is an example of a 5 by 4 grid with an X in the position B3.

A B C D E

1

2

3

4

Each player uses two grids. Each player uses one of their grids to arrange their ships and record the torpedoes fired by the opponent. On the other grid, the player records their own shots and whether they hit or missed.

Before play begins, each player secretly arranges their ships on their primary grid. Each ship occupies a certain number of consecutive squares on the grid (sizes of ships are in the following table), arranged either horizontally or vertically. The number of squares for each ship is determined by the type of the ship. The ships cannot overlap so only one ship can occupy any given square in the grid. The types and numbers of ships allowed are the same for each player.

The game is played in rounds. In each round, each player takes a turn to fire a torpedo at a target square in the opponent’s grid. The opponent then indicates whether the shot was a hit (a ship occupied the square) or a miss (there was not ship in the square). If the shot is a “miss”, the player marks their primary grid with a white peg (X in our game); if a “hit” they mark this on their own primary grid with a red peg (O in our game). The attacking player then indicates the hit or miss on their own “tracking” grid with the appropriate color peg (red (0) for “hit”, white (X) for “miss”) so that they can understand where the opponent’s ship might be.

In the board game, once all of the coordinates of a ship have been hit, the ship is sunk, and the ship’s owner announces “You sunk my battleship! (Or whatever the particular ship that was destroyed). For our purposes, we will consider a battleship sunk if the opponent has a single hit. When all of one player’s ships are sunk, the other player wins the game.

For your game, you will create a one-person version of the game where ‘the computer’ will play for the second player.

At the beginning of the game, you will read a file called ship_placement.csv which contains the type of ship, the first grid square for the ship placement, and whether the ship is placed vertically or horizontally (V or H in the field). The file will be in csv format (comma separated values). This is a common format and is comma separated (instead of being on separate lines). There will be commas between the values. Blank values will just have a comma noting to go to the next field (the game input should not have blank fields so you should handle the case where a field is blank). If you want to view the file, often this will be opened by a spreadsheet unless you specifically open it with a text editor. Do not open it with Microsoft Word, as this may change the format. The first line of a CSV file notes the data descriptions as follows:

TypeOfShip,Location,HorizOrVert

I have provided several sample files which contain good scenarios and scenarios with placement issues that you will need to handle using exception handling. Your game should run with any of these files, but should also be able to run with any valid file in the correct format. You will need to check whether all ships were included in the input file (and appropriate action to take if not), whether all ships have been placed, whether they fit on the board in the configuration given, and whether more than one ship occupies a space (which is not allowed) when you read the input file from the user and how to recover if an error occurs.

You will then need to randomly position the computer’s ships on the grid taking into consideration the same factors as you did for the player’s input.

You will need to prompt for and allow for the user to input their next guess in the form of a letter (A through J) and a number (1 – 10) indicating where they are targeting for their torpedo and you should error check the input. In our simplified game, you will determine if the torpedo shot was a hit or a miss. If the shot was a hit, consider the ship to be sunk. You should display a hit or miss, whether the ship was sunk and which one, and display their tracking grid so they know what they have guessed and where they have made hits. The entire ship which was hit will display as sunk.

After the user takes their turn, you must have the computer randomly select a shot that they have not previously taken. Then you must display to the user what the computer guessed, whether it hit any of the player’s ships, whether a ship was sunk, and then display the player’s placement grid showing where ships are located and what has been hit.

You should continue this until someone wins or quits the game – meaning you should allow the player to gracefully quit at any turn.

At the end of the game, you should indicate the game is over and who the winner was. You should also allow the user to quit the game by entering a Q when prompted for their next guess. If a player decides to quit the game, the grid with all of their guesses and the locations of the computer’s ships should be displayed.

Overall System Design

1. You must have two different classes in your design.

2. You must use inheritance in one of the classes.

3. When reading from a data file, your program should test the input file to ensure that data is of valid format (basic error detection) using Exception Handling.

4. You should consider using the Grids from Assignment 2 to make this easier. You do not need to have 4 grids for this but if you decide to use only two grids, you need to make sure you do not show the player the computer’s ship location when you display the grid after each turn.

5. Each component of the overall program should be modular.

1. Program should be fairly fault tolerant of user input and the appropriate user prompts and on-screen directions should be displayed

1. Split the program into multiple files based on the roughly categorized functionality or classes.

Extra Challenge / Extra Credit

Some of you may want an extra challenge to boost your abilities and have some interesting resume material. If you have completed all of the requirements above and want an additional challenge, you can incorporate fully sinking the ships and additional logic to be used by the computer when it makes a hit. After determining whether the shot was a hit or miss, you will then need to determine if the ship was sunk. In the simplified version of the game above, one torpedo hit sunk the ship. For the extra credit, you will need to accurately track all hits on a ship to determine if it was sunk. If it was sunk, you will need to tell the player, “you sunk my XXXship” where XXX indicates which type of ship and then displays the information as described in the simplified version above. Additionally, you will need to add logic to the random selection of torpedo shots for the computer. When the computer takes a random shot that hits a ship, the next series of shots will need to be logically selected until the hit ship is sunk.

If you do the extra credit, please note this in your documentation and make it clear in the running of the program, so we can give you up to 15% extra credit. (meaning you could get 115% on the project). Note: Since this is extra credit, it needs to meet a higher standard for full extra credit).

1

Do you need a similar assignment done for you from scratch? Order now!
Use Discount Code "Newclient" for a 15% Discount!

QUESTION 1

1. The ways that HLOOKUP formula works is searching down instead of across data. True
   False

0.9 points  

QUESTION 2

1. The AND formula takes a list of Booleans and returns TRUE if all of them are true, and FALSE otherwise. True
   False

0.9 points  

QUESTION 3

1. How would Excel evaluate the following formula? =OR(7<5, A2=6,NOT(“zebra”<=”tree”))? Assume that the formula is in cell D4, the workbook contains no circular references, and outside of cell D4 the workbook contains no errors.TRUEFALSEIt would return an errorThere is insufficient information to answer this question

1.4 points  

QUESTION 4

1. One of the benefits of using the Table feature is that you’ll be able to see the column titles at every moment. True
   False

0.9 points  

QUESTION 5

1. Use the brewery.xlsx file to answer the following question: How many total transactions took place at Beerland Amherst?3527425844258542194315939

0.9 points  

QUESTION 6

1. Use the file Code_Enforcement_-_Building_and_Property_Violations.xlsx for this question.
   What is the total dollar amount of penalties for closed cases in Charlestown due to residents’ improper storage of trash (improper storage trash:res)?

1.4 points  

QUESTION 7

1. Use the Brewery.xlsx file to answer the following question: What s the name of the beer with the highest Alcohol by Volume (ABV)?Half LordGear ShiftSheep MojoRed Salty HoserGnu Breath

0.9 points  

QUESTION 8

1. The PivotTable functionality is located in the Data tab. True
   False

0.9 points  

QUESTION 9

1. Use the Brewery.xlsx file to answer the following question: How many beers product names have order quantities between 3000 and 12000?75689

1.4 points  

QUESTION 10

1. Which of the following functions returns a range of cells?OFFSETCOUNTAMATCHREDEFINENone of the given choices

0.9 points  

QUESTION 11

1. A named range can be usedwhen creating chartsin formulasin functionsnone of the given choicesall of the given choices

0.9 points  

QUESTION 12

1. Which of the following statements is correct if new data will be attached as a new row to the existing table?OFFSET( base cell, 0, 0, 1, COUNTA (larger range in row where potential data may appear) )OFFSET( base cell, 0, 0, COUNTA (larger range in row where potential data may appear),1)OFFSET( base cell, COUNTA (larger range in row where potential data may appear), 0, 0,1)OFFSET( base cell, 0, COUNTA (larger range in row where potential data may appear), 0,1)None of the given choices

1.4 points  

QUESTION 13

1. One of the primarily characteristics of Solver is the mathematical relationship between the objective, the constraints, and the decision variables. True
   False

0.9 points  

QUESTION 14

1. The Solver and Analysis Toolpak Add-ins are not built in to Excel. True
   False

0.9 points  

QUESTION 15

1. Use solver to answer the following question: A manufacturing concern wishes to minimize its shipping costs between its 4 factories and 3 warehouses. Factory 1 supplies 1000 units per week and costs $5, $3, and $4 to ship each unit to Warehouses 1, 2, and 3 respectively. Factory 2 supplies 1200 units each week and costs $4, $3, and $3 to ship to Warehouses 1, 2, and 3. Factory 3 supplies 1500 units and costs $6, $2, and $5 to ship to the three warehouses. Factory 4 supplies 1800 units and costs $6, $2, and $4.
   If Warehouse 1 requires 3000 units per week, Warehouse 2 demands 1000, and Warehouse 3 demands 1500, what is the minimum it would cost them in shipping to fulfill each warehouse’s demand?

Do you need a similar assignment done for you from scratch? Order now!
Use Discount Code "Newclient" for a 15% Discount!

Do you need a similar assignment done for you from scratch? Order now!
Use Discount Code "Newclient" for a 15% Discount!