Scribd Logo
Upload

Welcome to Scribd!

  • 53 views

    Beercase

    Uploaded by

    tumeti
    The document provides a problem statement for a sales manager who wants to build a collaborative recommendation system for an online beer store to recommend beers customers are most likely to buy based on customer ratings data. It outlines several steps for solving the case study, including preparing the data by selecting beers with a minimum number of reviews, exploring the data to understand user and beer similarities and ratings, and developing recommendation models using item-based and user-based collaborative filtering and comparing their performance.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd

    Beercase

    Uploaded by

    tumeti
    53 views2 pages
    The document provides a problem statement for a sales manager who wants to build a collaborative recommendation system for an online beer store to recommend beers customers are most likely to buy based on customer ratings data. It outlines several steps for solving the case study, including preparing the data by selecting beers with a minimum number of reviews, exploring the data to understand user and beer similarities and ratings, and developing recommendation models using item-based and user-based collaborative filtering and comparing their performance.

    Original Title

    Beercase.docx

    Copyright

    © © All Rights Reserved

    Available Formats

    DOCX, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document provides a problem statement for a sales manager who wants to build a collaborative recommendation system for an online beer store to recommend beers customers are most likely to buy based on customer ratings data. It outlines several steps for solving the case study, including preparing the data by selecting beers with a minimum number of reviews, exploring the data to understand user and beer similarities and ratings, and developing recommendation models using item-based and user-based collaborative filtering and comparing their performance.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd
    Download as docx, pdf, or txt
    53 views2 pages

    Beercase

    Uploaded by

    tumeti
    The document provides a problem statement for a sales manager who wants to build a collaborative recommendation system for an online beer store to recommend beers customers are most likely to buy based on customer ratings data. It outlines several steps for solving the case study, including preparing the data by selecting beers with a minimum number of reviews, exploring the data to understand user and beer similarities and ratings, and developing recommendation models using item-based and user-based collaborative filtering and comparing their performance.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd
    Download as docx, pdf, or txt
    You are on page 1of 2

    Problem Statement

    You are the sales manager for "BeerMart", an online beer store in the United States.  You want to
    build a recommendation system (collaborative) for your store, where customers will be
    recommended the beer that they are most likely to buy. You collected the data about the ratings that
    the customers have provided in the past. You can download the dataset from the link below.

    Description: Each record is composed of a beer's name, the name of the user along with ratings
    provided by users. All ratings are on a scale from 1 to 5 with 5 being the best.

    As you solve the case study, answer the following questions

    1. Data preparation
    1. Choose only those beers that have at least N number of reviews
     Figure out an appropriate value of N using EDA; this may not have one correct
    answer, but you shouldn't choose beers having extremely low number of
    ratings
    2. Convert this data frame to a “realratingMatrix” before you build your collaborative
    filtering models
    2. Data Exploration
    1. Determine how similar the first ten users are with each other and visualise it
    2. Compute and visualise the similarity between the first 10 beers
    3. What are the unique values of ratings?
    4. Visualise the rating values and notice:
     The average beer ratings
     The average user ratings
     The average number of ratings given to the beers
     The average number of ratings given by the users
    3. Recommendation Models
    1. Divide your data into training and testing datasets
     Experiment with 'split' and 'cross-validation' evaluation schemes
    2. Build IBCF and UBCF models
    3. Compare the performance of the two models and suggest the one that should be
    deployed
     Plot the ROC curves for UBCF and IBCF and compare them
    4.  Give the names of the top 5 beers that you would recommend to the users "cokes",
    "genog" & "giblet"
    Rubrics

    Your assignment will be evaluated using the following guidelines.

    Step Meets expectations Does not meet expectations

    The data is in incorrect format;


    The data has been converted to
    beers have not been selected on
    Data preparation realRatingMatrix; beers having at least N
    the basis of minimum number
    ratings have been chosen to build the model
    of ratings

    Visualisations have been created to understand Visualisations have not been


    the similarity between first 10 beers and users; created; the average ratings and
    Data exploration
    the average ratings and number of ratings for number of ratings of beers and
    beers and users have been reported/visualised users are not reported or plotted

    Has experimented with split and cross-


    Not experimented with the two
    Recommendation validation methods; UBCF & IBCF models
    methods; ROC curve not
    models have been built and compared using ROC
    created
    curve or any other relevant metric

    You might also like