660 likes | 873 Views
Vinod Gupta School of Management, IIT Kharagpur. Google Refine Tutorial. April, 08 2012 Sathishwaran.R - 10BM60079 Vijaya Prabhu - 10BM60097. This Tutorial was created using Google Refine Version 2.5 on a Windows 7 platform. Data Cleansing.
E N D
Vinod Gupta School of Management, IIT Kharagpur Google RefineTutorial April, 08 2012 Sathishwaran.R - 10BM60079 VijayaPrabhu - 10BM60097 This Tutorial was created using Google Refine Version 2.5 on a Windows 7 platform
Data Cleansing • Data cleansing is identifying the wrong or inaccurate records in the data set and making appropriate corrections to the records. • It involves identifying incomplete, inaccurate, and incorrect parts of data and then either replacing them with correct data or deleting the incorrect data • Data cleansing results in data which is consistent with the other standard data and is useful for performing various analysis • The error in the data could be due to data entry error by the user, failure during transmission of data or improper data definitions.
Need for Data Cleansing • Incorrect or inaccurate data may lead to false conclusions and can cause investments to be misdirected in finance. • Also government needs accurate data on population and census for directing the funds to the deserving areas. • Many organizations tap into customer information. If the data is not accurate, for eg. If the address is not accurate then the business runs the risk of send wrong information, thus losing customers.
Challenges Data Cleansing • Loss of Information: In many cases the record may be incomplete, hence the whole record may require to be deleted which leads to loss of information. It could become costly if huge number of data is deleted. • Maintenance of Data: Once the data is cleansed then any change in the data specification needs to affect only the new values. Hence data management solutions should be designed in such a way that the process of data entry and retrieval are altered to provide correct data. • Data cleansing is an iterative process which needs significant work in exploration and corrction of entries.
About Google Refine • Google Refine is a powerful tool that can be effectively used for data cleansing. • It helps in working with raw data, cleaning it up, transforming from one format to other, encompassing it with web services and linking it to databases. • It is very easy to use and has a web interface. • It is freely available and works well with any browser. • Google Refine is a desktop application and it runs a small web server on your system and we need to point our browser to the server to use refine.
Getting Started - Installation • Download the zip file (appropriate Windows, Mac, Linux versions) from the link http://code.google.com/p/google-refine/wiki/Downloads?tm=2 • Uncompress the files from the zip file. • Run the “google-refine.exe” file. • A command window opens and Google refine runs taking the user to the home page in the default browser.
Importing Data • Google Refine supports TSV, CSV, Excel (.xls and .xlsx), JSON, XML, and Google data document formats. • Once imported the data is in Google Refine’s own data format. • We have used TSV data on Disasters worldwide from 1900-2008 available from http://www.infochimps.com/datasets/disasters-worldwide-from-1900-2008 for the tutorial.
Creating Project Data Uploaded
Creating Project Project Created
Faceting • Faceting is about seeing the big picture and filtering based on rows to work on data you want to change in bulk. • We can create a facet for a column to get the details about that column and then we can filter to a subset of rows with a constraint. • We can perform text facet, Numeric facet, timeline facet and scatterplot facet. Also various customized facets can be designed.
Faceting The Column Type has 18 unique options
Removing Redundancy Even though they are of same type, shows as different options due to case
Removing Redundancy Reduced to 15 unique options
Numeric Faceting Highly clustered towards low values
Numeric Faceting Cost column is blank and has no value
Numeric Faceting Calamities with low cost
Numeric Faceting Calamities with high cost
Clustering • Clustering is used to merge choices which look similar.
Clustering Data Merged
Using Expressions • Expressions are used to transform existing data to create new data
Data Augmentation • Reconciliation option in Google refine allows data to be linked to web pages. Suppose we want details on the country where the calamity has struck we can perform the following steps
How to Use Twitter Data Step 1 Step 2
Step 4 Step 5