The-Data-Wrangling-Report-

Introduction

Data wrangling is a core skill that everyone who works with data should be familiar with since so much of the world's data is not clean. The data wrangling process

• Gathering data.
• Assessing data.
• Cleaning data.

Gathering Data

Enhanced Twitter Archive

The Twitter archive #WeRateDogs, a csv file that contains the tweet, tweet id, timestamp, text, rating numerator and denominator, dog name, etc.

Image Predictions File

Image prediction file, based breed of dog is in each tweet, I downloaded the image prediction file programmatically from Udacity's servers using the requests library.

Additional Data via the Twitter API

Additional data collection including "Retweet count "and "Favorite count", using python's tweepy library.

Assessing Data

After gathering the data and storing them in DataFrames, assessing the data for quality and tidiness.

Data were assessed based on quality and tidiness.

• Low quality is data has content issues such as missing, inaccurate. So I doing I removing unnecessary columns, converting data types, and removing outliers.
• Untidy is data has structural issues. So I doing gathering dog stages from multiple columns into one, creating a “prediction” column (dog, not dog, maybe dog), and combining the three datasets into one.