Skip to content

Latest commit

 

History

History
105 lines (76 loc) · 3.78 KB

README.md

File metadata and controls

105 lines (76 loc) · 3.78 KB
Raw

Data Scraper by David Chen

This data scraper was implemented with python. This repo does two things: scrape data from bestbuy.com and do some light statistical and quantitative analysis. Python has great libraries such as BeautifulSoup and numpy that help solve these problems.

Scraping Data

In order to run the program, make sure BeautifulSoup and numpy is installed.

  • Enter this command on the Command Prompt to scrape: python run_scrape.py
  • This will scrape bestbuy.com with the search terms curved_smart_tv and smart_tv, filtered by Samsung, LG, Toshiba, Sony
  • The csv file will be saved in the data folder

The naming convention of the csv files will be in MMDDYY so that reading the data from a start date to end date will be simpler.

Reading Data

  • Enter this command on the Command Prompt to read the data: python read_data.py
  • The results that are printed out in the Command Prompt will aggregate ALL the days
  • If you need to look at only one day, create a new ReadData in read_data.py passing in a start date and None as the end date:
one_day = ReadData('091617', None)
  • If you need to look at a specified time period, create a new ReadData in read_data.py passing in a start date and end date:
two_days = ReadData('091417','091517')
  • Passing no parameters into ReadData will default to analyzing the entire data folder:
all_days = ReadData()

Data Model

An example data model:

{
    id: [ranking, search_term, brand, rating, reviews],
     0: ['1', 'smart_tv', 'LG', '4.6', '274'],
     1: ['2', 'smart_tv', 'Samsung', '4.6', '5899'],
     2: ['3', 'smart_tv', 'LG', '4.6', '187'],
     3: ['4', 'smart_tv', 'Samsung', '4.6', '3318'],
     4: ['5', 'smart_tv', 'Samsung', '4.7', '219'],
     5: ['6', 'smart_tv', 'Samsung', '4.6', '2538'],
     6: ['7', 'smart_tv', 'LG', '4.8', '51'],
     7: ['8', 'smart_tv', 'LG', '4.6', '231'],
     8: ['9', 'smart_tv', 'LG', '4.5', '443'],
     9: ['10', 'smart_tv', 'Samsung', '4.7', '360'],
     10: ['11', 'smart_tv', 'LG', '4.5', '130'],
     11: ['12', 'smart_tv', 'Samsung', '4.6', '2052'],
     12: ['13', 'smart_tv', 'Samsung', '4.5', '99']
 }

Answering Questions

Given a set of competitor brands, for a given time range, for a given search term, what percentage of search results are owned by each brand?
  • Pass in the relevant search_term into the brand_ownership class method
  all_days.brand_ownership('smart_tv')
  all_days.brand_ownership('curved_smart_tv')

From 09/14/17 to 09/17/17, the % of brand ownership for smart_tv: LG: 32%, Samsung: 65%, Sony: 3%.

From 09/14/17 to 09/17/17, the % of brand ownership for curved_smart_tv: Samsung: 100%.

Given a set of competitor brands, for a given time range, for a given search terms, what percentage of the top 3 search results are owned by each brand?
  • Pass in the relevant search_term into the top_brands class method
  all_days.top_brands('smart_tv')
  all_days.top_brands('curved_smart_tv')

From 09/14/17 to 09/17/17, the % of top 3 search results for each brand for smart_tv: LG: 67%, Samsung: 33%.

From 09/14/17 to 09/17/17, the % of top 3 search results for each brand for curved_smart_tv: Samsung: 100%.

Is there a correlation between the # of reviews and search ranking?
  all_days.review_vs_ranking()

With a correlation coefficient of -0.00057, the # of reviews and search ranking appear to be independent of each other.

Is there a correlation between the rating and search ranking?
  all_days.rating_vs_ranking()

With a correlation coefficient of -0.05871, there appears to be no correlation between rating and search ranking.

To view all of the answers that was run and saved:
  all_days.view_results()