Tweetl

By using Tweetl, you can simplify the steps from getting tweets to pre-processing them. If you don't have twitter API key, you can get it here.

This package help you to ・・・

• get tweets with the target name and any keywords.
• pre-processes the following list.
  • remove hashtags, URLs, pictographs, mentions, image strings and RT.
  • unify characters (uppercase to lowercase, halfwidth forms to fullwidth forms).
  • replace number to zero.
  • remove duplicates (because they might be RT.)

Installation

pip install Tweetl

Usage

Getting Tweets

Create an instance of the 'GetTweet' Class.

import Tweetl

# your api keys
consumer_api_key = "xxxxxxxxx"
consumer_api_secret_key = "xxxxxxxxx"
access_token = "xxxxxxxxx"
access_token_secret = "xxxxxxxxx"

# create an instance
tweet_getter = Tweetl.GetTweet(
                    consumer_api_key,
                    consumer_api_secret_key, 
                    access_token, 
                    access_token_secret
                )

With target name

You can collect tweets of the target if you use 'get_tweets_target' method and set the target's name not inclueded '@'. Then it returns collected tweets as DataFrame type. And you can specify the number of tweets.

# get 1000 tweets of @Deepblue_ts
df_target = tweet_getter.get_tweets_target("Deepblue_ts", 1000)
df_target.head()

With any keywords

You can also get tweets about any keywords if you use 'get_tweets_keyword' method and set any one. And you can specify the number of tweets.

# get 1000 tweets about 'deep learning'
df_keyword = tweet_getter.get_tweets_keyword("deep learning", 1000)

Cleansing Tweets

Create an instance of the 'CleansingTweets' Class. And using 'cleansing_df' method, you can pre-processing tweets. You can select columns that you want to cleanse. The default is only text-colmn.

# create an instance
tweet_cleanser = Tweetl.CleansingTweets()
cols = ["text", "user_description"]
df_clean = tweet_cleanser.cleansing_df(df_keyword, subset_cols=cols)

Author

deepblue

License

This software is released under the MIT License, see LICENSE.